Tag: Authority

Automated E-E-A-T Auditing: An Evidence-Led Workflow

Your crawler can find a missing byline in seconds. It cannot tell you, by itself, whether a reader should trust a consequential claim or whether Google will consider its creator authoritative. That distinction determines whether automated E-E-A-T auditing becomes a useful quality-control system or confidence theater.

A reliable audit collects observable evidence, judges that evidence against the purpose of each page, and sends uncertain or consequential decisions to a person. It turns a broad quality framework into a repeatable editorial queue without pretending that E-E-A-T is a metric you can retrieve from Google.

An automated audit finds evidence; it does not measure Google

E-E-A-T stands for Experience, Expertise, Authoritativeness, and Trustworthiness. Google uses it as a framework for evaluating content quality and credibility, but its guidance is not exposed through a simple API endpoint. Your tool therefore cannot request an official E-E-A-T score. Any percentage, grade, or traffic-light rating it produces is a summary of your own rubric.

That does not make automation useless. It changes what the tool should claim to do. A defensible auditor identifies evidence that a reviewer would use when making an E-E-A-T assessment:

For experience, it can locate descriptions of a process, first-hand observations, original methods, demonstrations, limitations, and outcomes. It cannot prove that the claimed experience happened.
For expertise, it can inspect bylines, biographies, qualifications, professional roles, explanatory depth, and support for factual claims. It cannot infer genuine expertise merely because the prose sounds confident.
For authoritativeness, it can connect a page to an identifiable creator or organization and find evidence of relevant work or recognition. An on-site crawl alone cannot establish the wider reputation of that entity.
For trustworthiness, it can check ownership, contact routes, dates, citations, disclosures, policies, corrections information, and consistency between visible content and structured data. It cannot verify every claim simply because the page contains references.

The right verdict vocabulary reflects those limits. Use labels such as observed, missing, ambiguous, not applicable, and not assessed. A failed browser request must produce not assessed, not missing. A weakly relevant biography should be ambiguous, not automatically accepted as expertise.

This distinction protects your editorial team from a common failure: treating a detector’s confidence as evidence of the underlying fact. The detector may be highly confident that it found a credential. Whether the credential is real, current, and relevant is a separate judgment.

Build a page-type-aware rubric before choosing a model

Three different page types are paired with distinct sets of evidence symbols and evaluation frameworks.

A universal checklist will punish pages for failing to be something they were never meant to be. A contact page does not need an expert byline. An author profile should not be judged as though it were a commercial landing page. An editorial policy can describe a review process, but its existence does not prove that the process was followed on every URL.

Start by classifying pages according to purpose. Then decide which evidence is applicable to each class. The following matrix is a practical starting point, not an official Google scoring model.

Page type	Primary audit questions	Misreading to prevent
Informational content	Who is responsible for the claims? Is relevant expertise or experience visible? Are factual assertions supported and limitations explained?	Treating fluent, detailed prose as proof of expertise.
Author or reviewer profile	Is the person identifiable? Are qualifications, roles, experience, and published work relevant to the subjects they cover?	Awarding expertise for a generic biography or an unrelated credential.
Homepage or about page	Who owns the site? What does the organization do? Is its purpose, identity, and relevant competence clear?	Counting promotional language as independent evidence of authority.
Commercial or service page	Is the seller identifiable? Are important claims substantiated? Can a customer find material terms, support, and an accountable contact route?	Assuming conversion copy is sufficient evidence of trust.
Editorial, disclosure, or corrections page	Are review, correction, sourcing, and commercial-disclosure processes explained clearly enough to be followed?	Assuming that a published policy proves consistent implementation.

Write each rubric check as an operational rule. Name the page types to which it applies, the evidence the auditor may accept, evidence that is insufficient, the allowed verdicts, the reason the check matters, and the remediation that follows a failure. If two reviewers cannot apply a rule consistently, an AI model will not rescue it.

For example, a rule called author expertise present is too loose. A better rule asks whether the page identifies its primary creator and whether the linked profile contains experience, qualifications, or work relevant to that page’s subject. The tool should return the creator’s name, the relevant evidence it found, the URL or element containing that evidence, and any ambiguity. It should not award expertise simply because an Author field exists in JSON-LD.

Structured data is valuable evidence about how a site represents its entities. It is not a substitute for the underlying reality. Compare author names, organization names, publication dates, review dates, and canonical URLs in markup with what a visitor can see. Flag contradictions as trust issues. Do not award credibility merely because the markup is syntactically complete.

Do not begin with a whole-site score. Begin with representative page types because one page cannot support a meaningful assessment of an entire website, while a complete crawl is often unnecessary during rubric development. Include the templates that publish important claims, the pages that establish creator and organization identity, and the governance pages those templates rely on. Expand only after the rules work on that sample.

Run a browser-based evidence pipeline

Abstract web pages move through an automated evidence pipeline while linked source items reach a human review station.

The model should be one component of the auditor, not the entire auditor. Retrieval, rendering, classification, deterministic checks, language-model judgment, and reporting solve different problems. Keeping them separate makes failures visible and lets you improve one layer without rewriting everything.

Define the audit unit. Record the site or section, locale, content types, excluded areas, and whether the run is a template sample or a broader crawl. This prevents results from unrelated markets or subdomains from being combined accidentally.
Inventory and classify URLs. Group pages by purpose and template before sampling. Classification can begin with URL patterns, metadata, headings, structured-data types, and internal-link context, but uncertain classifications should remain reviewable.
Select representative pages. Cover each important content purpose and template. Include identity and governance pages that provide context for individual URLs. A sample made only from high-traffic articles will miss the pages that establish who publishes the content and how it is controlled.
Render the pages. Basic fetchers can be blocked or can miss client-rendered content. A headless Chromium browser driven through Python automation can acquire the page as a browser sees it. Chromium and Selenium are practical examples, not requirements.
Extract evidence into a structured record. Capture the final URL, page title, headings, visible byline, linked profiles, visible dates, citations, policy links, contact details, relevant disclosures, internal and external links, and JSON-LD. Preserve where each item appeared rather than flattening the page into an unattributed text blob.
Run deterministic checks first. Code is better than an LLM at confirming that an element exists, a link resolves, a byline points to a profile, or visible and structured names disagree. Use language-model judgment for questions that require interpreting relevance, specificity, or context.
Apply the rubric with constrained outputs. Give the model the page class, the applicable criteria, the extracted evidence, and the allowed verdict labels. Require evidence for every observed or ambiguous result. Instruct it not to infer facts that are absent and not to penalize criteria marked not applicable.
Aggregate only after page-level review. Keep template patterns, page-specific findings, acquisition failures, and site-level context separate. A footer link repeated across every URL is one site-wide element, not fresh evidence on every page.

The acquisition status belongs in every result. Record successful rendering separately from blocked requests, authentication barriers, timeouts, parsing failures, unsupported files, and deliberate exclusions. Otherwise a crawler defect can generate a site-wide wave of false missing-evidence findings.

Keep the AI’s task narrow. It can judge whether a biography appears relevant to a subject, whether a passage describes a specific method, or whether a citation plausibly supports the nearby assertion. A human should decide whether credentials are authentic, whether high-consequence claims are correct, whether claimed experience is genuine, and whether external reputation supports an authority judgment.

Make every finding traceable and reviewable

An editor should be able to challenge an audit result without rerunning the entire system or reverse-engineering a prompt. Each finding needs a compact evidence trail:

The criterion and the page type that made it applicable.
The audited URL and acquisition status.
The verdict and confidence in that verdict.
The exact evidence used, kept to the shortest useful fragment.
The evidence location, such as a heading, link target, structured-data property, or DOM selector.
The rule or model version that produced the result.
A plain-language explanation of why the evidence passed, failed, or remained ambiguous.
A specific next action and the person or team best placed to take it.

Keep coverage separate from quality. If the auditor reached only part of the intended sample, report incomplete coverage prominently. Do not let the successfully audited pages create an apparently healthy site score while blocked or unclassified URLs disappear from the denominator.

A single composite score usually hides the decision an editor needs to make. Prefer an evidence matrix that shows status by criterion and page type, plus severity based on the consequence of the issue. A missing optional biography detail should not cancel out an identity conflict or an unsupported consequential claim merely because both affect the same average.

Controls for predictable failure modes

Retrieval failure looks like missing content. Gate all content judgments on successful acquisition and rendering.
Template elements inflate the result. Deduplicate repeated headers, footers, and policy links, then distinguish site-wide evidence from page-local evidence.
The model fills gaps with plausible assumptions. Require a captured evidence fragment and location for every positive verdict. Unsupported conclusions fail validation.
A generic checklist creates irrelevant failures. Mark applicability before scoring and retain not applicable as a real result.
Structured data earns unmerited credit. Treat markup as a claim about an entity, compare it with visible content, and flag mismatches instead of assuming truth.
An overall grade conceals serious findings. Report coverage, evidence status, ambiguity, and issue severity independently.
Prompt changes move the benchmark. Version the rubric, prompts, extraction logic, and result schema together. Re-run the validation set whenever one changes.
Stored page copies create avoidable content risk. Retain short evidence fragments, URLs, locations, and hashes where practical instead of archiving full third-party pages in the project repository.

Validate the auditor before expanding the crawl

Create a human-reviewed set of representative pages and record the expected applicability, evidence, verdict, and rationale for each check. Compare the automated output with those decisions. Inspect false positives and false negatives by criterion rather than celebrating agreement at the report level. A system that reliably finds bylines may still be poor at judging whether qualifications are relevant.

Test uncomfortable cases deliberately: a credential that is impressive but unrelated, a methodology paragraph with no indication that the creator performed the work, a policy that exists but is not linked from relevant pages, conflicting author names in visible content and JSON-LD, and a browser failure that leaves the extracted body empty. These cases reveal whether the auditor follows evidence or merely rewards familiar patterns.

Keep the rubric, prompts, test cases, and extraction code in version control. A project can begin inside an AI coding environment for flexible, multi-session iteration, or become a standalone application deployed outside that environment. The first shape suits a rubric that is still changing. The second becomes useful when you need repeatable runs, controlled access, scheduled processing, and a stable interface. Deployment does not make the judgments more valid; validation does.

Human review should remain visible in the final report. Record whether a finding is machine-only, reviewer-confirmed, changed by a reviewer, or awaiting specialist verification. Those states let you measure where automation saves time and where it still creates work.

Key takeaways

An automated E-E-A-T audit measures evidence against your rubric; it does not retrieve a Google score.
Classify pages by purpose before applying checks. Applicability is part of the judgment, not an afterthought.
Use browser rendering for acquisition, deterministic rules for objective checks, and an LLM only where interpretation is required.
Require every verdict to point to captured evidence and its location. Unsupported positive findings are as dangerous as false warnings.
Report acquisition coverage, ambiguity, and severity separately instead of compressing everything into one grade.
Validate on human-reviewed edge cases, version the whole system, and expand the crawl only when the findings lead to sound editorial decisions.

Start with one important page template and the identity or policy pages that support it. Label a representative set by hand, define what acceptable evidence looks like, and make the auditor explain every verdict. If it cannot distinguish absent evidence from inaccessible evidence, or observation from inference, it is not ready to scale. Once reviewers can turn its findings into precise edits without redoing the audit themselves, add the next template.

References

Search Engine Land – How to build an E-E-A-T checker with Claude Code

August 21, 2026

Multi-Location SEO Page Architecture That Scales Cleanly

Your location URLs keep multiplying, but rankings, calls and visits are not. Launching another city page may look like the quickest way to reach a new market, yet excess geographic pages can make your own URLs compete, divide authority and contradict one another.

A durable architecture works in the opposite direction. You represent the places where the business actually operates, give every page a distinct customer job and publish the smallest set of geographic URLs that can do those jobs well. Here is how to design that system, evaluate proposed city pages and clean up an existing footprint without discarding useful local information.

Map the operating footprint before choosing URLs

Hands arrange branch, service-area and customer markers on an unlabeled layered regional map.

Start with the business, not a keyword export. Build a working inventory of facilities, teams, services and markets before deciding what belongs under /locations/. This prevents a common category error: treating every place name as evidence of a separate local entity.

Your inventory should record:

Every customer-facing facility, including its official name, address, hours and primary contact path.
The staff or team responsible for each facility and market.
The services actually available at each location, rather than the complete company-wide service list.
The regions used operationally by the business, such as states, metro areas or franchise territories.
The communities each facility or field team can genuinely serve.
Material local differences, including access, logistics, regulations, delivery conditions or customer procedures.
The person or system responsible for keeping each local fact accurate.

Then classify each geographic concept. A physical facility, a regional market, a service area and a city the company wants to rank in are not interchangeable.

Operating reality	Customer need	Default architectural response
Customer-facing facility	Confirm where it is, when it is open, what it offers and what visiting involves	Create an authoritative location page
Region containing multiple facilities	Understand the brand’s presence and choose the appropriate facility	Create a regional hub only when it materially helps that choice
Service area reached by a facility or field team	Confirm coverage and understand how service is delivered	Explain it on the responsible location or service page unless the market has enough distinct substance for an exception
City the business wants to rank in	Discover a relevant provider	Treat it as a marketing objective, not an automatic page type

Service-area settings in Google Business Profile should not determine this map. Adding a city to a profile does not require a city landing page, and publishing a page does not create a physical presence there. The website must remain honest about whether customers visit you, you travel to them, or both.

At the end of this exercise, every proposed page should point back to an operating fact. If all you can point to is search volume, you have found a keyword opportunity, not yet a reason for a new URL.

Build a hub-and-spoke system around customer decisions

Most multi-location sites need a central locations directory connected to regional or individual location pages. The depth depends on the business. A larger network might use /locations/, /locations/pennsylvania/ and /locations/pennsylvania/philadelphia/. A smaller regional company might need only /locations/ and /locations/philadelphia-pa/. Neither folder pattern is inherently more optimized; the useful pattern is the one that mirrors the real hierarchy without inserting empty layers.

The main locations hub helps people orient themselves

The hub should explain the overall footprint and help a visitor reach the right facility. A map, postcode search or location finder can improve the experience, but it should complement a crawlable directory rather than replace it. Include direct links to important regional and location pages so people and crawlers can navigate the footprint without operating an interactive widget.

Organize that directory in the way customers choose: by region, proximity, service availability or another real decision factor. Do not add state and city levels merely to make the URL look comprehensive.

Regional hubs resolve a choice between facilities

A regional page earns its place when it helps someone understand a meaningful market or compare several facilities. It can describe the coverage model, identify available locations, clarify material differences and send the visitor to the correct next page.

A region with only a heading, generic brand copy and links to a single destination is an unnecessary layer. Link the main hub directly to the location unless the regional URL has a durable job of its own.

Location pages represent real facilities

A location page is more than an organic landing page. It is the business’s authoritative digital representation of that facility. Someone arriving from search, navigation, an AI answer or a shared link should be able to confirm that the place is real and decide what to do next.

Include the local facts that change the decision:

Official location name, address, contact details and opening hours.
Services available at that facility, with links to the relevant service pages.
Local staff or team information when it helps customers know whom they will deal with.
Directions, arrival instructions and recognizable local context.
Parking, entrances, mobility access and other accessibility details.
What happens after the visitor calls, books or arrives.
A conversion action appropriate to that facility, such as calling, booking, requesting service or getting directions.

Do not manufacture superficial rewrites merely to achieve an arbitrary uniqueness percentage. Accurate service descriptions, brand language and booking instructions may need to recur. The decisive question is not whether some copy is shared, but whether the page has a distinct reason to exist. Its differentiation should come from local reality, not a thesaurus.

Service and location pages answer different questions

A service page explains what the company offers. A location page explains where and how customers receive it. Keep both roles intact and connect them deliberately:

From a location page, link only to services genuinely available there.
From a service page, help the customer find the facilities or teams that provide it.
From a regional hub, link to the facilities contained in that market.
From the main hub, expose the regional or location pages that form the real operating hierarchy.

A service-area page is a controlled exception within this system. It may be justified when the market has a dedicated team, distinct logistics, local regulatory conditions or substantial project experience that cannot be handled properly on an existing page. Willingness to drive into a city is not enough.

Make every proposed geographic page pass an evidence test

Keyword demand can reveal an audience, but it cannot tell you whether that audience needs a separate destination. Before approving a geographic page, require the requester to answer these questions in writing:

What customer task will this page complete? The answer should be more specific than ranking for a city term.
What real operation does it represent? Name the facility, team, territory, logistics model or other business fact behind it.
Why can’t an existing page satisfy the same intent? Identify the gap instead of assuming a new URL is the cure.
Which facts are genuinely local? Look for distinct staff, services, access, regulations, logistics, projects or customer expectations.
Does it lead to a meaningful local action? The conversion path should match how the business serves that market.
Where does it belong in the hierarchy? Define its parent page and the service, regional or location pages that should link to it.
Who will maintain it? A page containing hours, services or team details needs an accountable owner.
Would its purpose survive if you removed the city name from the draft? If nothing substantive remains, you probably have a keyword variant rather than a useful page.

The physical-location question carries the clearest answer: a real customer-facing facility generally warrants a location page. A service-area proposal needs stronger operational evidence because the place name alone does not represent a separate entity.

Consider a field team that leaves from one facility and serves surrounding communities with the same staff, services, process and booking path. A separate page for every community would mostly change the city name while funneling every visitor to the same operation. The better answer is usually one strong facility or service page that clearly explains its coverage.

Now consider a market with its own team, different delivery constraints, local rules and a body of market-specific work. That page can answer questions the parent location page cannot. It has an operational identity and a customer job, not merely a keyword.

This distinction also keeps the site away from a doorway-like pattern. Pages become risky when they target closely related queries, offer little market-specific value and send visitors toward the same destination. Not every weak city page constitutes doorway abuse, but a large collection of near-identical funnels is poor architecture even before policy becomes the concern.

Consolidate geographic bloat without erasing useful local value

A maze of similar doorways merges into a central hall leading to a few distinct local spaces.

Geographic sprawl usually accumulates through individually plausible decisions: a city-keyword project, neighborhood pages around a branch, a franchise microsite or a replacement URL structure that leaves the old one intact. The result is often an architecture that no team fully owns.

Do not begin the cleanup by changing folders or deleting low-traffic pages. Begin with a complete URL inventory and group pages by the intent they satisfy, the operation they represent and the conversion destination they use.

Find every geographic URL. Combine CMS exports, XML sitemaps, crawl data, navigation links and known campaign landing pages. Include orphaned pages that are still indexable even if they no longer appear in menus.
Record evidence before making changes. Capture each page’s business entity, target intent, organic landing activity, conversions, internal links, external links and current indexation status. This keeps a quiet but useful customer page from being mistaken for dead weight.
Cluster overlapping pages. Put URLs together when they answer the same geographic query, represent the same facility or team, and send visitors to the same conversion path. Similar titles alone are not enough; compare the job each page performs.
Assign a disposition. Keep a page with a clear, durable job. Merge pages whose useful information belongs on one authoritative destination. Repurpose a page only when a genuine uncovered customer need exists. Retire a URL that has no distinct entity, intent or maintained value.
Select the surviving destination by utility. The winner should best represent the real operation and satisfy the visitor, even if another duplicate happens to have the preferred slug. Traffic is evidence to consider, not a substitute for architectural logic.
Preserve worthwhile local information. Move accurate directions, accessibility details, team information, service availability or project context to the surviving page before retiring a duplicate.
Redirect deliberately. When content has a relevant replacement, use a permanent redirect to that destination. Do not send every retired city URL to the homepage; that breaks the geographic intent instead of resolving it.
Update the system around the URL. Change internal links, navigation, directory listings, canonical references and XML sitemaps so they point directly to the surviving page rather than through a redirect.
Verify the result. Crawl the revised section, test important customer paths and watch indexation, landing-page activity and conversions for unexpected losses or lingering duplicate URLs.

A page should not be removed merely because it attracts little organic traffic. Location pages also help customers verify a facility, understand the visit and take action. If the page serves that role well, improve its discoverability and local facts rather than judging it as a failed keyword landing page.

Add governance so the bloat does not return

A cleaner tree will expand again unless page creation has an owner and an approval rule. Use a short request record for every new geographic URL. It should name the page type, operating entity, customer job, parent page, market-specific evidence, conversion path and maintenance owner.

Maintain one dependable business-data record for addresses, hours, contacts, services and local ownership. Templates can then reuse stable brand and service information while pulling the local facts that make each facility accurate. This is more valuable than asking writers to disguise duplication with cosmetic wording changes.

When the business opens, closes, relocates or changes what a facility offers, update that record and its dependent pages as one operational task. Architecture is not finished when URLs launch; it succeeds when the site can remain correct as the footprint changes.

Key takeaways

Build the location tree from facilities, teams, services and real markets before using keyword demand to refine it.
Treat physical locations, regional markets, service areas and desired ranking cities as different concepts.
Use regional hubs only when they help customers understand a market or choose among multiple facilities.
Make each location page the authoritative customer resource for its facility, including services, hours, staff, directions, access and next steps.
Approve service-area pages only when distinct operations or market-specific information give them a durable customer purpose.
Consolidate pages that satisfy the same intent and lead to the same operation, then redirect and update internal signals deliberately.
Require a business owner and maintenance plan for every geographic URL.

If you take one action this week, freeze new city-page requests long enough to build the operating-footprint matrix. Place every current and proposed URL beside the facility, region, team or service condition that justifies it. The blank rows will show you where keyword ambition has outrun business reality.

Start cleanup with the clearest overlap, preserve the information customers still need and give the surviving page a single accountable owner. A leaner location system will not manufacture local relevance, but it will make the relevance you genuinely have easier for customers, search engines and AI retrieval systems to understand.

References

Search Engine Land – Multi-location SEO: How to structure geographic pages at scale

July 30, 2026

Leading SEO and GEO Practitioners in 2026: A Field Guide

If you are deciding whom to follow, invite into a strategy session, or hire in 2026, a generic “top expert” list will not solve the real problem. The person who can untangle multilingual crawling may not be the right person to build AI citation visibility, and the clearest interpreter of Google policy may not offer client services at all.

Use this field guide to route your problem to the right kind of practitioner. It separates public authority from specialist fit, advisory insight from delivery capacity, and conventional SEO expertise from the newer work required across ChatGPT, Claude, Gemini, Perplexity, and other generative interfaces.

A useful shortlist is a map, not a podium

SEO and GEO now overlap, but they are not interchangeable. SEO generally improves discoverability, relevance, and performance in conventional search results. GEO focuses on whether a brand, product, or expert is accurately represented, cited, or recommended in generative answers. AEO sits across both, especially where content must supply a concise answer that a search feature or AI system can extract.

A leading practitioner therefore needs to be leading in relation to a particular job. Technical architecture, international deployment, algorithm recovery, industry reporting, content authority, entity clarity, AI citation measurement, and lead generation require different combinations of experience. Treating them as one discipline produces impressive-looking shortlists and weak hiring decisions.

Public prominence is useful evidence, but it is not proof of fit. Keynote history supplies 35% of one 2026 expert-scoring model; books carry 20%, citations 15%, and tenure, active blogging, and social reach 10% each. That formula measures contribution, recognition, and audience more directly than it measures implementation quality, client continuity, or business outcomes.

One material conflict also deserves your attention. Evan Bailyn is First Page Sage’s president, while First Page Sage assigns the top position to Bailyn and to its own agency. That makes those placements self-rankings. They can identify a credible candidate, but they should not replace independent references, attributable results, or a close examination of who will actually perform the work.

Key takeaways

For an SEO and GEO program tied to B2B lead generation, start with Evan Bailyn, but independently validate the claims made by his own firm.
For multilingual or multiregional SEO, Aleyda Solis has the clearest specialist fit.
For technical architecture and development, consider Jono Alderson; for internal linking and content scoring, study Cyrus Shepard’s work, although he is listed as unavailable for hire.
For site-quality or algorithm problems, Marie Haynes and Lily Ray are better starting points than a generalist. Barry Schwartz is more useful for monitoring what changed.
For Google policy and search history, follow Danny Sullivan for context, not consulting; he is listed as unavailable for hire.

Match each practitioner to the problem in front of you

Fictional specialists examine separate models representing multilingual, technical, local, content, and AI search problems around a strategy table.

The following map is intentionally problem-first. Availability reflects the cited 2026 information and can change, so confirm it before building an outreach plan.

Practitioner	Best fit	Listed for hire in 2026?	What you should verify
Evan Bailyn	Thought-leadership SEO, GEO, and lead generation	Yes	Independent outcomes, named involvement, and how AI visibility connects to qualified demand
Aleyda Solis	International, multilingual, and multiregional SEO	Yes	Experience with your markets, languages, architecture, and implementation constraints
Barry Schwartz	SEO news and Google algorithm-update monitoring	Yes	Whether you need reporting, diagnosis, or implementation; these are different deliverables
Marie Haynes	Site quality, algorithm updates, and penalty recovery	Yes	Evidence distinguishing an update impact from technical failure, demand change, or competition
Jono Alderson	Technical SEO and web development	Yes	Implementation ownership, engineering access, and the handoff from diagnosis to shipped changes
Lily Ray	Algorithm analysis, search quality, AI, and organic search	Yes	Which work belongs to SEO versus GEO and how each stream will be measured
Cyrus Shepard	Technical SEO, internal linking, and content scoring	No	Current availability and whether his published frameworks can be implemented by your team
Danny Sullivan	Google search policy, algorithm communication, and SEO history	No	Use his work for policy context rather than treating it as account-specific advice

SEO and GEO tied to lead generation

Among these names, Bailyn is positioned most explicitly at the intersection of SEO, GEO, thought leadership, and lead generation. The associated enterprise practice focuses on content authority, third-party validation, and entity optimization intended to improve brand representation in AI-generated answers. That combination is relevant when your buyers conduct long, research-heavy evaluations and may encounter an AI-generated recommendation before reaching your site.

The important question is not whether those workstreams sound reasonable. It is how they connect. Ask which audience questions will be monitored, which AI interfaces will be tested, what sources currently shape the answers, what assets will be changed, and which commercial action should follow improved visibility. A growing citation count is an intermediate signal; it is not revenue evidence by itself.

International and technical SEO

Solis is the more precise choice when your difficulty crosses languages, countries, or regional site structures. Her work covers multilingual crawl analysis and international architecture, while her SEOFOMO newsletter also tracks developments in AI search. Before hiring any international specialist, provide a market-by-market inventory. Include domains or subdirectories, languages, local publishing ownership, shared templates, and the markets that matter commercially. Without that inventory, even a strong practitioner has to spend the opening phase discovering the shape of the assignment.

Alderson and Shepard occupy a more technical lane, but they are not identical choices. Alderson’s combination of technical SEO and web development is useful when recommendations must survive contact with an engineering backlog. Shepard’s stated specialties make him especially relevant to internal linking and content scoring. If your immediate need is a repeatable backlink process or training for an internal marketing team, Brian Dean is an additional specialist to consider. None of these briefs is equivalent to owning a full enterprise GEO program.

Quality, algorithms, and the search news cycle

Schwartz, Haynes, Ray, and Sullivan help at different moments. Schwartz is the monitoring layer: use his work to learn that a change, test, or industry development is occurring. Haynes is a closer match when rankings or traffic have fallen and site quality or a Google update may be involved. Ray bridges search-quality analysis with AI and organic search. Sullivan’s three decades in search and his 2017-2025 period as Google’s public Search Liaison make him important for policy context and historical interpretation, but he is not a consulting option.

Do not ask a news specialist to prove the cause of your decline merely because they reported the update first. Start with the timeline, affected directories, query groups, page types, conversions, technical changes, and competitive movement. Then choose the practitioner whose specialty matches the remaining uncertainty.

A public expert and a delivery team are different purchases

Following a practitioner gives you ideas, vocabulary, and early warning. Hiring a practitioner should give you accountable decisions. Hiring an agency should also give you production capacity, measurement, project management, and continuity. Those are three different purchases, even when the same name appears in all of them.

The enterprise GEO market illustrates the available operating models:

First Page Sage describes a high-touch, founder-led model built around thought leadership, SEO, GEO, authority, and entity optimization. If senior involvement is important, put the expected involvement in writing rather than relying on the sales process.
Genevate, established in 2025, was built as a GEO-first firm. Its work includes AI citation audits, benchmarking, authority-led content, and a proprietary citation dashboard. The specialization is attractive, but its short operating history leaves less evidence about long, complex enterprise programs.
Driven Metrics, also established in 2025, emphasizes analytics, attribution, and real-time citation tracking across ChatGPT, Perplexity, and Gemini. Its enterprise portfolio is narrower than those of longer-established firms, so test its capacity against your number of markets, products, stakeholders, and approval layers.
NP Digital combines GEO with SEO, paid media, and content through a global team. That breadth can simplify multi-channel management. Client feedback summarized for 2026 also raises the risks of account-team turnover and reduced senior-strategist involvement after setup, making continuity an important diligence question.
Terakeet, established in 2004, brings a longer enterprise history in organic marketing, brand authority, narrative control, and reputation. Seer Interactive, established in 2002, is another longer-tenured option with a data-driven SEO and GEO orientation.

A dashboard should not decide this choice for you. Citation tracking can reveal whether selected prompts produce your brand, competitors, or supporting sources, but the result depends on the prompt set, model, interface, timing, location, language, and method of repetition. Ask to see the measurement specification, not just the dashboard screen.

Your agreement should identify who owns strategy, who attends recurring reviews, who approves content, who handles technical recommendations, and who explains a material performance change. If you are buying access to a named practitioner, specify that person’s role. If you are buying a delivery system, assess the system instead of assuming the public figure will supervise every decision.

Run this diligence before you hire an SEO or GEO expert

An evaluation team reviews technical models, project materials, and delivery capacity during a meeting with a fictional search consultant.

You do not need a sprawling request for proposal to distinguish a specialist from a polished seller. A tightly framed problem and a consistent set of questions will tell you more.

Define the failure in one sentence. Name the affected asset, audience, market, and outcome. “We need GEO” is not a usable brief. “Our product is absent when North American procurement leaders ask AI assistants to compare vendors in our category” gives a practitioner something concrete to investigate.
Ask for competing explanations. A credible candidate should be able to distinguish crawl or indexation problems, weak relevance, inadequate authority, poor entity clarity, reputation issues, demand changes, and measurement errors. Immediate certainty before access to evidence is a warning sign.
Make the candidate draw the SEO-AEO-GEO boundary. Ask which recommendations improve conventional search, which improve extractable answers, and which are intended to influence generative representation. Shared tactics are normal. Pretending the three labels mean exactly the same thing is not.
Inspect the measurement design. For SEO, look for a dated baseline covering visibility, indexation, qualified organic visits, conversions, and relevant business outcomes. For GEO, request the prompt portfolio, models and interfaces tested, languages or regions, repetition method, citation and mention rules, answer-accuracy checks, and downstream behavior where it can be measured.
Trace one complete evidence chain. Ask for a prior example that connects baseline, diagnosis, intervention, changed search or AI behavior, and business consequence. Redacted evidence is acceptable. A logo slide, an isolated screenshot, or a percentage without its denominator is not the same thing.
Confirm ownership and capacity. Identify the people doing discovery, analysis, content review, technical work, executive communication, and weekly decisions. Then ask how many accounts those people support and what happens if the lead strategist leaves.
Check references that resemble your assignment. A famous client name proves little if your challenge involves more regions, a regulated review process, a different buying cycle, or a larger implementation burden. Ask references about the work performed, the people who remained involved, the evidence delivered, and the problems that were not solved.

A five-part scorecard for the final decision

Score each candidate from zero to two on five dimensions: problem fit, verifiable evidence, measurement quality, delivery ownership, and honest treatment of constraints. Zero means absent or unsupported, one means plausible but incomplete, and two means specific and verifiable. Do not let a strong total conceal a zero for evidence or ownership. Those gaps usually surface after the contract is signed, when changing providers is more costly.

Promises that should stop the conversation

A guarantee that a particular model will cite or recommend your brand.
A GEO plan consisting only of adding schema or rewriting pages for AI. Structured data can clarify machine-readable facts, but it does not create third-party authority or guarantee inclusion in a generated answer.
AI share-of-voice numbers without a stable prompt set and documented test method.
Performance screenshots without dates, baselines, comparison periods, or definitions.
A sales process led by a recognized practitioner with no contractual explanation of that person’s delivery role.
A claim that mentions or citations are automatically equivalent to qualified traffic, pipeline, or revenue.

Build a roster that does not depend on one guru

If your immediate goal is to follow the field, assign each person a job. Schwartz can monitor the news cycle. Sullivan can supply policy and historical context. Haynes and Ray can sharpen your thinking about quality and algorithm effects. Alderson and Shepard can anchor technical questions. Solis can cover international architecture. Bailyn can contribute the SEO-to-GEO and lead-generation perspective, with the self-ranking caveat kept visible.

You do not need to follow every voice equally. When something changes, start with the monitor, move to the relevant specialist, and test the interpretation against your own site or AI-visibility data. This prevents a fast industry opinion from turning into an expensive implementation before the cause is understood.

Your next step is small: write one sentence naming the failure, asset, market, and desired outcome. Send the same brief to two appropriately matched specialists and score their responses on fit, evidence, measurement, ownership, and constraints. The leading practitioner for you is the one who reduces the right uncertainty and connects the work to a result your organization actually values.

References

July 24, 2026

Semantics and Topical Authority: A Local SEO Framework

You have a list of services, locations, modifiers, and customer questions. The uncomfortable part is deciding whether that list should become 20 useful pages or 200 repetitive ones. Publish too little and you leave real search needs unanswered. Publish every possible combination and your pages begin competing with one another.

The solution is a semantic page map: one in which every URL owns a distinct search task, related questions are covered at the right level, and local variations exist only when they give the searcher a meaningfully different answer.

Topical authority begins with clean page boundaries

Publishing volume is not topical authority. A site can have thousands of pages and still make its subject harder to retrieve because several URLs offer roughly the same answer. A smaller site can be more coherent when each page has a clear purpose and the pages collectively cover the important parts of the topic.

Semantics helps you draw those boundaries. Instead of treating a query as a bag of keywords, break it into the entities, attributes, relationships, context, and action that create its meaning. Consider a query such as emergency water heater repair in Oakville. Its semantic components include:

Service entity: repair.
Equipment entity: water heater.
Attribute: emergency or urgent availability.
Geographic context: Oakville.
Likely task: determine whether help is available and take the next step.

Those components do not automatically justify four or five pages. Emergency may belong on the main repair page. Oakville may need a location page, or it may belong in a service-area section. The equipment type may justify a separate page if it changes the service, the answer, and the search demand. Semantics gives you the ingredients; page planning decides how they should be represented.

A useful planning heuristic frames topical authority through historical performance, topical coverage, retrieval cost, and visual semantics. Treat that as an editorial model, not a published Google ranking formula. Its practical lesson is sound: broad coverage is valuable only when a search system and a visitor can efficiently find the right answer.

Some overlap is necessary. A furnace repair page and an air-conditioning repair page may both explain booking, diagnosis, and service areas. That shared material establishes their relationship and supports useful internal links. The problem begins when the pages have the same primary answer, structure, examples, and conversion path, with only the equipment or city name changed.

Use Query Deserves a Page before creating a URL

Hands sort small service, location, and question objects into a few separate circular trays on a desk.

Query Deserves a Page, or QDP, asks whether a search need should receive its own indexable URL. It replaces the reflexive question Can we publish this page? with the more useful question Does the searcher need a distinct destination?

The framework uses four signals for separating query variations:

Search demand: enough people look for the query family to make a dedicated destination useful.
Different entities: the query concerns a meaningfully different service, product, condition, audience, or place entity.
Low similarity: satisfying one query would not adequately satisfy the other.
A repeatable pattern: the query belongs to a recognizable template that appears across related entities or attributes.

None of these is a universal numeric cutoff. Use them as combined evidence. High demand alone does not rescue a page whose answer duplicates another URL. A different entity alone may not justify a page if the visitor receives the same information and takes the same action. Conversely, a lower-volume query may still deserve a page when it represents a clearly different service and a materially different decision.

What changes between queries	Likely representation	Planning decision
The primary entity and the answer both change	Dedicated page	Give the page one clear query family and a distinct purpose.
An attribute changes, but the main task remains the same	Heading and section	Answer the variation inside the strongest parent page.
The visitor needs a compact comparison or qualification check	Table, list, form, or information card	Make the answer visible without fragmenting it into another URL.
Only a browsing dimension changes	Filter or navigation control	Help users narrow choices without indexing every possible combination.

This distinction matters because a query can deserve representation without deserving a page. A pricing question might need a substantial section. A service-area question might need a concise list or lookup form. A product attribute might work best as a filter. Creating a URL for every answer is how useful coverage turns into index clutter.

Use this sequence for each proposed page:

Normalize the query family. Group close verbal variations together instead of treating every wording as a separate topic.
Name the primary entity. Identify the service, product, problem, audience, or location the searcher is actually trying to understand.
Write the search task in one sentence. For example: confirm that urgent water heater repair is available in Oakville and learn how to request it.
Compare it with the nearest existing page. Ask whether the entity, necessary answer, or next action changes enough to justify another destination.
Choose the smallest sufficient representation. Decide between a page, section, paragraph, list, table, form, card, or filter.
Assign one owner. Record which URL will be the primary destination for the query family.

When the evidence is ambiguous, consolidation is the safer starting point. A strong page can be divided later when a distinct search task becomes clear. Recovering several overlapping pages is usually more complicated because links, internal anchors, content, and performance history have already been distributed among them.

Build the topical map around entities and query templates

A three-dimensional network connects generic storefront, home, tool, tree, parcel, and location-marker objects in orderly branches.

A keyword list tells you what phrases exist. A topical map tells you how the underlying subjects relate and where each answer belongs. For local SEO, build that map along two axes: entity coverage and query-template coverage.

Entity coverage processes the meaningful attributes of the business and its subject. Depending on the business, that can include services, products, problems, equipment, customer groups, delivery methods, decision criteria, and genuine geographic relationships. Query-template coverage processes the recurring ways people ask about those entities, such as:

[service] in [location]
emergency [service] in [location]
cost of [service] in [location]
does [symptom] require [service]
[entity A] versus [entity B]
can [condition] be repaired

The brackets describe patterns, not a publishing quota. If you offer three services in ten locations, the matrix contains 30 service-location combinations before you add any attributes or questions. That does not mean 30 pages deserve to exist. Run every cell through QDP and select the right representation.

Create a working sheet with one row per query family and these fields:

Canonical entity: the main subject that should remain stable across verbal variations.
Context: local service, research, comparison, urgent help, qualification, or another genuine task.
Attributes: the modifiers that alter what the searcher needs.
Geography: city, district, neighborhood, service area, or no local modifier.
Query template: the repeatable form represented by the family.
Demand evidence: the data you use to show that the query family exists.
Current owner: the URL already receiving or targeting the family.
Planned representation: page, section, list, table, form, card, or filter.
Parent relationship: the hub or broader page that should link to this answer.

This sheet exposes two common planning errors. The first is a coverage gap: an important entity or question has no owner. The second is ownership duplication: several URLs claim the same query family. Both deserve attention, but they require opposite actions. Fill a real gap; consolidate an artificial split.

Your site architecture should then reflect those relationships. A broad service hub can introduce the service family. Dedicated pages can handle services that pass QDP. Qualified location pages can address place-specific tasks. Supporting resources can answer diagnostic, cost, comparison, and process questions at the depth they deserve.

Internal links should explain the relationship, not merely distribute authority. Link from the parent service to a specialized service when the reader may need that narrower answer. Link from a qualified location page to the service it offers. Use anchors that describe the destination accurately. Avoid linking every location page to every other location page simply because the template makes it easy.

Create local pages only when place changes the answer

Location pages are where a sensible topical map most often becomes a page factory. A business serving many cities can generate hundreds of city-service combinations, but a changed place name does not necessarily create a changed search task.

Apply five gates before approving a local landing page:

Operational truth: the business genuinely offers the named service in the named place.
A distinct local task: people need a destination for that location-service relationship, not merely another mention of the city.
A changed answer: location affects useful details such as availability, service boundaries, access, fulfillment, branch contact, or the correct next step.
Enough substance: the page can answer the local task without relying on paragraphs copied from neighboring pages.
Ongoing accuracy: the business can maintain the local claims when operations change.

A simple editing test catches many weak pages. Replace the city name with a neighboring city. If the page remains equally accurate and useful without any other change, geography probably has not altered the answer. Consider a broader regional page, a service-area section, a location selector, or another consolidated representation.

Do not mistake cosmetic variation for semantic differentiation. Reordered sentences, swapped synonyms, altered headings, different stock images, and repeated structured data do not create a new page purpose. Structured data should describe the entities and relationships that are actually present; it cannot supply a distinction the visible page does not have.

Nor does every repeated sentence create a duplicate. Shared brand facts, service processes, and contact instructions may legitimately appear across related pages. Duplication is query-dependent: documents can be treated as unique for one query and near-duplicates for another because the decisive issue is how much their query-relevant content overlaps. Keep the common material where it helps, but make the primary answer unmistakably different.

For example, air-conditioning repair and furnace repair concern different equipment and can require different answers, so they are plausible page candidates. Emergency air-conditioning repair may instead be an attribute covered prominently on the main repair page unless it represents a genuinely distinct service and search task. Air-conditioning repair for two neighboring cities should not become separate pages merely because both city names appear in query data.

The same discipline applies outside home services. A law firm does not automatically need a page for every practice-area, accident-type, city, and district combination. A treatment provider does not automatically need a page for every condition, treatment, and country permutation. The number of combinations describes the size of the matrix, not the number of pages you should publish.

Audit query ownership before expanding topical coverage

Before commissioning more content, find out whether your current URLs already have clean ownership. New pages cannot repair an architecture in which several old pages compete for the same job.

Run the audit in this order:

Inventory indexable URLs. Include service pages, location pages, resources, product or category pages, and any programmatically generated variants.
Assign a primary query family. Write one owner statement for each URL. If you cannot describe its job in one sentence, the page boundary probably needs work.
Group competing owners. Flag cases where multiple URLs target the same entity, context, location, and action.
Compare query-relevant content. Ignore global navigation and standard brand copy. Compare the headings, central answer, lists, tables, calls to action, and structured data that matter to the target query.
Choose an action. Keep, consolidate, differentiate, or retire the URL based on whether a distinct search task exists.
Rebuild the internal-link path. Make the selected owner easy to reach from its parent and remove ambiguous links to competing destinations.
Measure by query template. Group performance for families such as [service] [city], emergency [service], and cost of [service], then watch which URL earns the relevant visibility and actions.

Micro-cannibalization is often subtle. You may not see two pages occupying the same results at once. Instead, different URLs can appear for close variations of the same query family, preventing any one page from becoming the stable owner. Your ownership sheet makes that fragmentation visible.

Use four actions consistently:

Keep: the URL has a clear entity, task, and place in the hierarchy.
Consolidate: two or more URLs satisfy essentially the same need. Preserve the strongest material in one owner and update the links that pointed to the alternatives.
Differentiate: a real second task exists, but the current pages fail to express the distinction. Rewrite the purpose, central answer, headings, and internal links around that task.
Retire: the page represents no useful search need and contributes nothing necessary to a stronger owner. Do not delete first; preserve useful material, map the surviving destination, plan the redirect, and update internal links.

Keep the map inside the business’s genuine subject. Expanding into loosely related topics may produce more URLs, but it weakens the relationship between the site’s entities and the audience’s real needs. The underlying model assumes that successful performance can help a site become more competitive across similar entity-context pairs and query templates, while irrelevant or lower-quality expansion can undo that advantage. Treat that as a strategic model rather than a guaranteed account of every ranking decision.

Key takeaways

Topical authority comes from coherent coverage and clean retrieval paths, not page count.
Every query may deserve an answer, but only some queries deserve a dedicated page.
Use demand, entity difference, query similarity, and repeatable patterns as combined QDP evidence.
Map entities and attributes alongside query templates so you can see both coverage gaps and competing owners.
Approve a location page only when the place-service relationship is real and geography changes the useful answer.
Consolidate overlapping URLs before expanding the site into additional services, questions, or locations.

Start with one important service family. List its entities, attributes, locations, and recurring query templates. Assign every family to one existing URL or one smaller on-page element before proposing anything new. The result will show you exactly where another page can earn its place—and where restraint will make the whole site stronger.

References

Search Engine Land — How semantics and topical authority improve local SEO

July 22, 2026

When Original Research Becomes an AI Citation Benchmark
Original research can give AI systems something unusually valuable: a defensible answer that does not exist on every competing page. Yet the available citation analysis suggests that publishing proprietary numbers is not enough. The strongest results appear when those numbers form a benchmark that resolves a specific comparison.

That distinction changes the content strategy. The goal is not merely to demonstrate that a company has data. It is to turn first-party evidence into a transparent, retrievable answer to a question buyers are already asking.

The citation advantage is substantial but concentrated

An analysis reported by Search Engine Land examined Gauge’s set of 301 live pages cited by AI systems across 316 unique prompts and seven verticals. Those pages collectively received 1,075 citations. Only eight pages, or 2.7% of the cited set, qualified as primary research under the analysis’s definition: they presented original data and explained its methodology.

Despite their scarcity, those eight pages accounted for 90 citations, or 8.4% of the total. They averaged 11.3 citations per page, compared with 3.4 for the other pages. On that measure, primary-research pages were approximately 3.3 times as citation-dense as pages without primary research.

The result supports a useful but limited conclusion. Within this cited-URL set, original research was associated with disproportionately high citation volume. It does not establish that any page containing proprietary data will earn citations, nor does it measure the success rate of all published research. The dataset begins with pages that had already been cited, so it reveals patterns within successful sources rather than the probability that a new study will succeed.

Concentration inside the research subset makes that qualification especially important. According to the same report, 75 of the 90 primary-research citations came from a cloud data warehouse benchmark cluster. A Fivetran warehouse benchmark received 44 citations by itself, while two Fivetran benchmark pages together accounted for 58 of the 90. Once that cluster was removed, original research had a much smaller presence in the citation set.

A benchmark gives proprietary data a clear job

The reported pattern is better understood as a benchmark advantage than a general research advantage. A benchmark measures named alternatives against a defined yardstick and publishes comparable results. It can therefore answer questions such as which product is faster, less expensive or more efficient under stated conditions.

This format aligns the evidence with the shape of a commercial query. When a prompt asks an AI system to compare options, a benchmark supplies entities, criteria and results in one source. A collection of interesting statistics may demonstrate expertise, but it is less useful if the numbers do not resolve a recognizable decision.

The warehouse examples illustrate that alignment. Search Engine Land reported that the primary-research citations clustered around prompts involving measurable characteristics such as speed, cost, latency, yield and performance. Fivetran, Estuary and ClickHouse had numerical evidence applicable to those comparisons. In the crypto and Solana area, Marinade and Helius received citations for firsthand data relevant to staking and MEV questions.

The pattern was not uniform across subjects. After the source’s data cleaning, no cited primary-research pages were found in its B2B SaaS and CRM, education and TEFL, or product analytics topics. Those areas instead surfaced formats such as explainers, product pages, case studies and listicles. This does not show that benchmarking is impossible in those markets. It indicates that the observed citation advantage appeared where the prompt, metric and competing entities could be connected cleanly.

Retrievability turns a study into citation infrastructure

The Fivetran example helps separate data creation from citation readiness. Its reported performance was not attributed to one isolated statistic. The page combined a direct comparison, visible methodology, supporting material and a structure that made individual answers easy to locate.

A bounded question and recognizable entities

The benchmark named BigQuery, Redshift, Snowflake and Databricks and evaluated them on speed and cost. This creates a close match between a buyer’s comparison and the content’s entities and measurements. The research is not simply about cloud infrastructure in general; it is organized around identifiable choices.

A method readers can inspect

Search Engine Land reported that Fivetran used actual customer usage rather than relying only on synthetic assumptions. The page explained the queried data, the queries used, and the configuration and tuning of each warehouse. It also linked to underlying data and supporting references. Those elements allow a reader to examine where the results came from and where comparisons might cease to be equivalent.

Limits, corrections and a stable home

The benchmark included dated correction notes from December 2022, qualitative limitations and a caveat about a performance floor. These disclosures narrow the claim instead of presenting the result as universal. The source also noted that the URL remained at one canonical address: a page published in 2022 was still receiving citations in the analyzed 2026 data.

Together, these features make the page function less like a campaign asset and more like durable reference material. Clear result headings help isolate relevant passages; methodology makes the figures interpretable; raw material supports verification; and corrections preserve trust without discarding the accumulated authority of the original URL.

Research planning should begin with the decision

A benchmark-oriented program starts by identifying a recurring question that can be answered with evidence the publisher is genuinely positioned to collect. The relevant opportunity is not necessarily the largest available dataset. It is the gap where buyers compare named alternatives but lack a credible, well-scoped source with reproducible measurements.

The metric must also represent the decision fairly. A speed comparison needs declared workloads and configurations; a cost comparison needs a consistent basis; and any ranking needs boundaries that prevent a conditional result from appearing universal. Methodological disclosure is therefore part of the product, not supporting material to add after publication.

Editorial structure matters for the same reason. A useful benchmark states the question, identifies the compared entities, defines the yardsticks, presents the result and explains why it may differ from other findings. Descriptive headings should connect each passage to a likely reader question. Supporting data, source notes, limitations and dated corrections should remain attached to the canonical page.

This approach also establishes a higher bar for deciding what deserves publication. Proprietary numbers that cannot support a meaningful comparison may still be useful for internal analysis, thought leadership or market education. They should not automatically be treated as citation assets. The observed advantage belongs to research whose evidence, question and presentation reinforce one another.

Key takeaways
- In the reported Gauge set, primary-research pages were rare but averaged about 3.3 times as many citations per page as other cited pages.
- Most primary-research citations were concentrated in cloud data warehouse benchmarks, so the result should not be generalized to every proprietary-data article.
- The strongest format compares named options using explicit, commercially relevant measurements.
- Methodology, underlying data, limitations, correction notes and a stable canonical URL help turn a result into a durable reference.
- A research brief should begin with the buyer’s decision and work backward to the data, metric and test conditions needed to answer it responsibly.
As more publishers produce original data, scarcity alone will become a weaker differentiator. The more durable opportunity is to build benchmarks that remain understandable, inspectable and useful whenever an AI system or a person needs to make the comparison again.

References
- Search Engine Land — Why My Original Data Gets Cited Only as Benchmarks
July 8, 2026
How Authority Signals Shape Visibility in AI Search
AI search visibility depends on more than whether an individual page is relevant. The systems producing recommendations, comparisons, and summaries may also need enough consistent evidence to understand the organization, product, or person behind that page.

The two source articles approach this challenge from different directions. One examines entity understanding through a Google patent; the other argues for differentiated content and co-citation analysis. Together, they suggest that authority is built through a recognizable identity, distinctive knowledge, and credible associations across the wider information environment.

AI visibility begins with a legible entity

The article about Google’s 2023 patent reports that a proposed system could use large language models to extract information from websites and public data, identify relationships, generate summaries, and develop what the patent describes as a deeper characterization of an entity. The source says the term can encompass people, businesses, places, objects, and concepts.

This matters because a conversational search system has a different task from a conventional document index. Finding a page that contains matching words is not the same as deciding which business belongs in a recommendation, which products can be compared, or which source can reliably explain a subject. Those tasks require some conception of identity: what the entity is, what it offers, which subjects it is associated with, and how its claims relate to information elsewhere.

A patent describes a possible method, not proof that every feature is operating in search exactly as written. Its practical value is therefore directional. It provides a useful model for auditing whether a brand leaves enough coherent evidence for an AI system to identify and characterize it without relying on a single optimized page.

Authority combines consistency with differentiation

Consistency helps systems connect references to the same entity, but consistency alone does not establish authority. A perfectly uniform digital footprint can still be generic, derivative, or unsupported.

The second source supplies the complementary argument. Its author reports being among a group of 25 invited by Google in May 2025 to discuss the evolution of search results pages at Google I/O. According to that account, the central message was to create non-commoditized content. Because the supplied article is incomplete, that report should not be stretched into a detailed description of Google’s ranking systems. It does, however, introduce an important editorial distinction: information that merely repeats the market consensus is less useful for establishing a source as uniquely valuable.

These perspectives address different failure modes. Inconsistent names, descriptions, offerings, and relationships can make an entity difficult to resolve. Undifferentiated content can make a clearly resolved entity easy to overlook. AI visibility therefore requires both identity clarity and information value.

Co-citation reveals the authority network around a brand

Co-citation analysis examines which entities or sources are mentioned together in relevant documents. Used as a strategic lens, it shifts attention from isolated backlinks or rankings to the network of associations surrounding a subject. The second source frames this type of analysis as a way to support stakeholder approval, while the patent-focused source emphasizes relationships as part of a broader entity characterization.

The synthesis is useful even without assuming a particular ranking mechanism. If recognized organizations, specialists, products, and concepts repeatedly appear together in credible discussions while one brand is absent, that absence exposes an authority gap. The response should not be to manufacture mentions. It should be to identify what the visible entities contribute that the missing brand does not yet demonstrate: original expertise, useful evidence, a distinct point of view, public relationships, or clear subject ownership.

Co-citation also helps separate identity problems from reputation problems. A brand may publish extensive content but use inconsistent descriptions across its website, social profiles, and third-party listings. Alternatively, it may be described consistently yet rarely appear in independent discussions of the category. The first condition calls for entity reconciliation; the second calls for stronger contributions and earned recognition.

Key takeaways
- Make the entity unambiguous: Align core names, descriptions, offerings, expertise, and relationships across owned profiles and public references.
- Publish information with a reason to exist: Add analysis, evidence, experience, or framing that cannot be replaced by a generic summary of existing pages.
- Audit associations, not just keywords: Examine which organizations, experts, products, and concepts appear together in credible category coverage, then identify meaningful gaps.
- Distinguish presence from authority: Repetition can reinforce identity, but independent recognition and differentiated knowledge make that identity more credible.
- Treat patents as directional evidence: Use the reported Google patent to inform strategy without presenting its proposed methods as confirmed production behavior.
Build an evidence trail that systems can interpret

A practical AI visibility program should connect editorial, technical, brand, and public-relations work around the same entity model. The website needs to state clearly who the organization is and what it knows. Content needs to demonstrate distinctive value. External coverage needs to provide genuine corroboration and relevant associations. Public profiles need to reinforce rather than contradict those signals.

The emerging objective is not to repeat a preferred description everywhere or chase citations as isolated trophies. It is to create a coherent, independently supported body of evidence from which search and AI systems can form a reliable understanding. Brands that make both their identity and their contribution easy to verify will be better positioned as AI-mediated discovery develops.

References
- CrushPress.AI — Teaching AI Who You Are: The New Frontier for SEO
- CrushPress.AI — Mastering Stakeholder Approval with Co-Citation Analysis
June 22, 2026
How to Build Content Authority Across AI Search Engines
Content authority in AI search is not a single score that a brand earns once and carries everywhere. The source reports point to a more conditional system: visibility depends on the AI engine, the topic and prompt, the sources retrieved, and whether a useful passage can be extracted from the page.

That changes the optimization task. Instead of producing one broad guide and accumulating undirected mentions, publishers need to decide where they want to appear, understand what that system retrieves for the topic, and create evidence-rich passages that can survive the final selection process.

Authority now operates at three distinct layers

Taken together, the sources suggest that AI visibility has three layers: engine selection, topical trust, and passage extractability. A weakness at any layer can prevent a brand from being cited even when its conventional search performance is strong.

The engine layer determines which index, retrieval process, and content formats are likely to enter consideration. Uncover 7 Unmissable AI Search Trends Transforming Marketing reports that ChatGPT and Claude shared only 8% of citations in the analysis it covered. It also reports substantial format differences: community sites accounted for about 16% of ChatGPT citations, while Claude cited listicles 36% of the time and opinion content 13.2% of the time, compared with approximately 20% and 7.2%, respectively, for ChatGPT.

The topic layer determines whose evidence the system treats as relevant and credible. Boosting AI Visibility: Mastering Topic-Driven Authority argues that citation sources cluster by subject rather than following one universal hierarchy. Its examples indicate that competitor domains had a larger role in invoicing queries than in starting-a-business queries. A publication that matters for one part of a market may therefore contribute little authority to an adjacent part.

The passage layer determines whether the system can isolate a clear answer from the selected document. Mastering AI Search: Building Machine-Friendly Content reports a 66% extraction rate for pages under 5,000 characters and 12% for pages over 20,000 characters. Those figures should be treated as findings reported by that article, not as a universal length rule. Their strategic significance is that authority without retrievable statements may never become a citation.

Choose the engine and prompt class before optimizing

An AI-search plan should begin with the audience and the engine it uses, not with a generic content calendar. The trends report says 64% of sites cited by Claude appeared in Google’s top 50 for corresponding queries, compared with 37% of sites cited by ChatGPT. It further reports that 79.2% of Claude citations aligned directly with the top 10 Brave Search results in the analysis it references. Within that reported environment, Brave rankings offer a more observable diagnostic for Claude than conventional Google rankings alone.

Audience context may also affect prioritization. Citing Ramp’s AI Index, the trends article reports Anthropic usage at 34.4% of businesses and OpenAI at 32.3%. It also says approximately 85% of Anthropic’s revenue came from enterprise and API usage. These figures do not establish that every B2B organization should optimize for Claude first, but they support testing Claude as a distinct business channel rather than treating its consumer web traffic as a complete measure of relevance.

Even a priority engine is not equally optimizable for every prompt. According to the same report, ChatGPT initiated web searches for nearly 95% of prompts in the cited analysis, while Claude did so about one-third of the time. Claude was reportedly more likely to search for current-event, ranking, location, and comparison prompts, with reported search rates of 81%, 67%, 55%, and 51%, respectively. Definitions and procedures were described as much less likely to trigger retrieval.

This distinction prevents a common measurement error. A page cannot win a fresh web citation when an engine answers from internal model knowledge without searching. Prompt testing should therefore record whether retrieval occurred before a team interprets a missing citation as a content or authority failure.

Query expansion adds another engine-specific variable. The trends report characterizes ChatGPT fan-out queries as changeable, while reporting that Claude produced the same fan-out strings 65% of the time and attached the current year to 94% of them, compared with 17% for ChatGPT. Stable expansions may support tightly targeted pages; volatile expansions call for broader coverage across owned, earned, community, and other relevant sources.

Design passages around problems, claims, and constraints

Machine-friendly content is not simply shorter content. The more useful objective is modularity: each section should resolve a recognizable subproblem without requiring an AI system to reconstruct the answer from a long narrative.

The machine-friendly content report recommends replacing broad category positioning with problem-specific positioning. Its illustrative shift is from identifying a company merely as an insurance provider to explaining that it addresses underwriting for first-time drivers under 25 who have been declined by standard insurers. The example also shows why constraints matter. Stating who a solution is not for, where it applies, or what condition changes the answer can make a claim more precise and credible.

Headings should name the outcome or question addressed by the section. Paragraphs should open with a direct answer or citable claim, then add conditions, evidence, and explanation. The same source reports that explicit headings increased retrieval likelihood by 17.54% and says Gemini may use approximately 380 words for query grounding. These reported limits reinforce the value of self-contained sections, although they do not justify stripping away evidence or necessary nuance.

The synthesis is a two-level editorial model. At page level, the article should offer a coherent argument for a human reader. At passage level, it should state entities, relationships, qualifications, and evidence clearly enough to be extracted independently. Narrative still has a role, but it should extend a usable answer rather than delay it.

Build off-site authority inside the relevant source network

On-site clarity makes a document usable; it does not make the publisher trusted by every system or for every topic. The topic-driven authority report recommends mapping the domains, publications, experts, and platforms that repeatedly appear in answers for the exact subject a brand wants to own. This is more focused than pursuing links or publicity from generally prominent sites without checking their topical role.

That mapping should also distinguish content formats. The topic-authority report describes YouTube as an exception that can surface across larger language models and recommends working with recognized subject-matter experts and relevant LinkedIn voices. The engine trends report, meanwhile, finds that community content was more prominent in ChatGPT citations and that listicles and opinion pieces were more prominent in Claude citations. Together, these observations suggest that the right distribution mix depends on both the topic’s trusted entities and the target engine’s retrieval preferences.

Concentration may matter more than raw mention volume. The authority report argues that recognition can move in jumps when a brand earns coverage from a highly trusted topical source, and it recommends ranking potential collaborators by authority tier. This remains a strategic recommendation from the source rather than proof that every high-profile placement will produce citations. Teams should validate it by comparing citation frequency before and after individual placements.

Measurement should follow the same conditional structure. For each priority prompt, a useful record includes the engine, whether it searched the web, the apparent query expansions, cited domains, cited passage types, the brand’s inclusion, and the presence of paid placements. The trends report says ChatGPT ads can appear around competitor mentions, so organic citation monitoring and paid competitive monitoring should be kept separate. Otherwise, a purchased appearance can be mistaken for earned authority, or a strong organic mention can obscure a competitor’s paid defense.

Key takeaways
- Define authority by engine and topic; citation strength in one model or subject does not automatically transfer to another.
- Confirm that the target prompt triggers web retrieval before investing in pages intended to earn fresh citations.
- Build problem-specific, self-contained sections with direct claims, explicit conditions, and enough evidence to stand alone.
- Concentrate outreach on the publications, experts, communities, and formats that already shape answers for the target topic.
- Measure retrieval, organic citations, and paid placements separately so each visibility mechanism can be diagnosed accurately.
As retrieval systems, source preferences, and advertising models change, durable advantage will come from maintaining this engine-topic-passage map as a living operating system rather than treating AI optimization as a one-time rewrite.

References
June 17, 2026
How Trust Turns Vehicle Shipping Interest Into Bookings
Vehicle shipping customers are often asked to commit before they can directly evaluate the service. That makes conversion less a matter of adding persuasion and more a matter of reducing uncertainty about price, responsibility, timing, vehicle handling, and communication.

The supplied First Page Sage article frames this relationship in its headline, How Trust Drives Conversions at AutoStar Transport Express. Its available excerpt identifies an interview with Mark Dugger, described as AutoStar Transport Express’s operations manager, but it does not provide enough detail to attribute particular tactics or results to the company. The useful lesson is therefore best developed as a broader conversion framework rather than an unsupported case study.

The conversion barrier is uncertainty, not simply price

A shipping quote gives a prospective customer a number, but the decision also depends on what that number appears to cover. A low price can lose persuasive value if the buyer cannot tell who will handle the vehicle, whether important conditions are excluded, or what happens when plans change.

This is the central connection between trust and conversion: trust makes an offer easier to evaluate. It does not require the customer to assume that every variable is predictable. Instead, it gives the customer a clear picture of which parts of the process are known, which may vary, who is accountable, and how changes will be communicated.

That distinction matters in vehicle shipping because operational complexity cannot always be removed from the service. The stronger conversion strategy is to explain complexity in language a buyer can use, rather than conceal it behind an apparently simple promise.

Trust signals should answer the buyer’s next question

Identity and responsibility: A prospective customer should be able to understand who the business is, what role it plays in arranging or providing transport, and where responsibility sits at each stage. Company information and credentials are most useful when they clarify accountability rather than merely decorate a page.

Quote clarity: The quote experience should explain inclusions, potential variables, payment expectations, and the conditions that could affect the final arrangement. Clarity is a trust signal because it helps buyers compare offers on substance instead of comparing headline prices that may not represent equivalent services.

Process visibility: Customers benefit from knowing what follows a request, how pickup and delivery are coordinated, what information they will receive, and whom they can contact. A visible process converts an abstract promise into a sequence the buyer can understand.

Evidence with context: Reviews, testimonials, and other forms of social proof are more informative when they address relevant concerns such as communication, issue handling, and whether expectations matched the delivered service. Evidence should support the operating claims on the page, not substitute for explaining them.

Realistic language: Absolute assurances can create suspicion when a service depends on changing operational conditions. Precise language about estimates, contingencies, and communication procedures can be more credible than an unqualified guarantee.

A trustworthy journey stays consistent from page to follow-up

Trust can be weakened when individual parts of the conversion journey contradict one another. An informative landing page does little good if the quote form introduces unexplained requirements, or if a follow-up message uses pressure that conflicts with the measured tone of the site.

The message should remain consistent across search results, service pages, quote forms, confirmation messages, phone conversations, and booking documents. The same terminology should describe the service and its conditions throughout. If a detail becomes more nuanced later in the journey, the earlier page should prepare the customer for that nuance.

Forms also communicate risk. Asking only for information needed at that stage, explaining why sensitive details are required, and showing what happens after submission can reduce hesitation. The immediate response should confirm receipt, set an appropriate expectation for the next contact, and preserve the claims that led the customer to inquire.

Operational delivery completes the conversion system. Marketing may secure the booking, but communication after booking determines whether the original trust claim remains credible. That experience can later influence reviews, recommendations, repeat business, and the evidence available to future customers.

Measure whether clarity changes customer behavior

A trust initiative should be tied to a defined point of uncertainty. For example, a business might clarify quote inclusions, explain its role in the transport process, make the next step more visible, or revise language that sounds more certain than the operation allows. Each change should have a reason grounded in customer questions or observed friction.

Quote completion and booking conversion can reveal whether more visitors progress, while abandonment points and recurring questions can show where uncertainty remains. Cancellation reasons, complaints, and mismatches between quoted expectations and later conversations provide a necessary counterweight: a higher initial conversion rate is not a success if it produces more misunderstanding afterward.

A/B testing can help distinguish the effect of a particular presentation change from normal variation, provided the test changes a clearly defined element and uses an appropriate measurement window. Qualitative feedback remains important because conversion data can show where behavior changed without explaining why.

Key takeaways
- Trust improves conversion by making the shipping offer easier to understand and evaluate.
- Useful trust signals answer concrete questions about identity, responsibility, quote scope, process, and communication.
- Credentials and reviews are strongest when they reinforce clear operating claims rather than stand alone.
- Realistic explanations of variables can be more credible than promises that remove all uncertainty.
- The full journey, from landing page through post-booking communication, should maintain the same expectations.
- Conversion gains should be assessed alongside cancellations, complaints, and expectation mismatches.
The next competitive advantage is likely to come from treating customer uncertainty as operational feedback. Businesses that connect recurring questions to clearer pages, forms, follow-up, and service communication can improve the booking experience without asking buyers to rely on persuasion alone.

References
- First Page Sage — How Trust Drives Conversions at AutoStar Transport Express
June 8, 2026
Google May 2026 Core Update: A Practical Recovery Plan
Your traffic graph dropped during the May core update, and now you need to know whether to rewrite pages, change your SEO strategy, or simply wait. Start by resisting the urge to make sitewide edits. A core update can expose weak content, but it can also coincide with changes in demand, search-result layouts, competitors, or tracking.

The useful response is a page-level diagnosis. You want to identify where visibility changed, determine what those pages now fail to deliver, and improve them without destroying content that still works.

Anchor your diagnosis to the actual rollout

The official rollout ran from May 21 through June 2. Noticeable ranking movement appeared by May 23 and continued into the following week. This was the second core update of 2026, so earlier changes in your reporting may belong to a different event.

Build clean comparison periods

In Google Search Console, compare May 7-20 with June 3-16. These are equal 14-day periods immediately before and after the rollout, without mixing rollout days into either side. If your business has strong weekly or seasonal patterns, compare each period with the equivalent days from a normal prior period as a second check.

Export clicks, impressions, click-through rate, and average position by query and page. A chart of total clicks is not enough. It can tell you that performance changed, but not why.

Separate ranking losses from other traffic losses

If positions declined across several important queries for the same pages, investigate relevance, usefulness, and competition. If impressions declined while positions remained broadly stable, check whether search demand or the set of queries triggering those pages changed. If positions and impressions held steady but click-through rate fell, inspect the live results for new answer features, stronger titles, or a changed search intent.

Also rule out unrelated technical problems. Check whether affected URLs are indexed, canonicalized as intended, crawlable, and returning the correct status code. Review analytics changes, security incidents, migrations, and major template releases. A core-update diagnosis cannot fix a broken canonical or missing tracking tag.

Find the losses that actually need intervention

Sitewide averages hide the decisions you need to make. Group affected URLs by topic, search intent, template, author, and content type. Then calculate the change for each group. A fall concentrated in old comparison pages calls for a different response than a decline across every page using the same template.

Start with URLs that combine three traits: a material visibility loss, meaningful business value, and a problem you can clearly describe. Do not prioritize a page merely because its percentage decline looks dramatic. A page that fell from ten impressions to two is usually less urgent than one that lost a large share of qualified visits.

Inspect the queries that disappeared

For each priority URL, compare its pre-update and later query sets. Ask whether it lost its main query, a cluster of secondary questions, or visibility for terms that never matched its real purpose. Losing poorly matched impressions may not require a repair. Losing the queries that express the page’s central promise does.

Search those important queries manually and examine the pages now appearing above yours. Look for differences in intent, scope, specificity, first-hand evidence, freshness, and format. The goal is not to copy competitors. It is to understand what searchers can accomplish with the current results that they cannot accomplish with yours.

Look for patterns across winners and losers

Your unaffected and improving pages are useful controls. Compare them with declining pages from the same site. If both groups share the same design, author box, and schema, those elements are less likely to explain the difference. If losses cluster around thin location pages, outdated tutorials, or articles built from the same generic outline, you have a stronger hypothesis to test.

Audit for satisfaction, not an imaginary update factor

The update was intended to favor relevant and satisfying content. That direction is more useful than hunting for a new word-count target, schema type, or keyword-density rule. Google has not provided a single prescribed fix for pages that lost visibility.

Test whether the page fulfills its promise

Read the title, opening, and major headings without relying on your memory of the page. They should define one clear task or question. Then check whether a reader can complete that task without returning to search for missing steps, definitions, evidence, or limitations.

Remove introductions that delay the answer. Put the central answer or decision criteria near the relevant heading, then support it with explanation. If the query requires a procedure, make the sequence explicit. If it requires a choice, explain who each option suits and what changes the decision.

Add value that another generic page cannot reproduce

A rewrite that merely changes wording preserves the original weakness. Add the missing substance: a worked example, a transparent method, a limitation, an expert interpretation, a screenshot that proves a step, or an explanation of what happens when the standard advice fails. Keep only material that helps the reader act or decide.

For factual or high-consequence claims, make the basis visible. Identify the responsible organization or expert where that identity matters. Link to supporting material you actually used. Show when the page was reviewed, and update that date only after a meaningful review. An unexplained assertion does not become trustworthy because it sounds confident.

Check ownership, duplication, and internal competition

Decide which URL should own each core intent. Several pages targeting the same question can divide internal links and leave each version incomplete. Consolidate genuine duplicates when one stronger destination can serve the reader. Keep separate pages when the intents, audiences, or required answers are materially different.

Update internal links so descriptive anchor text points to the intended owner. Make sure related pages support one another instead of repeating the same opening-level information. Do not delete a large group of URLs solely because traffic fell during the rollout; first determine whether each page has a distinct, supportable purpose.

Improve search and AI visibility without conflating them

A Google core update and visibility inside frontier language models are not the same measurement system. A decline in Google rankings does not prove that ChatGPT, Claude, or another answer engine stopped citing you for the same reason. Track conventional search performance and AI citations separately, even when the same content improvements may benefit both.

For answer-oriented visibility, make important facts easy to locate and interpret. Use descriptive headings, answer the stated question directly, name entities consistently, and keep qualifications beside the claim they modify. Tables should represent real comparisons, while lists should represent genuine steps or criteria. Formatting cannot compensate for an unsupported answer, but it can make a strong answer easier to extract correctly.

Apply JSON-LD only when it accurately describes visible content and the page’s real entity relationships. Schema is packaging, not evidence. Adding more markup will not repair stale facts, unclear authorship, duplicated intent, or an answer that misses the query.

Measure AI visibility with a stable set of prompts tied to your customers’ questions. Record whether your brand is mentioned, cited, represented accurately, or omitted. Keep that record beside, but not merged into, your Search Console analysis. This prevents a gain in one channel from concealing a loss in another.

Make controlled changes and preserve what you learn

Create a change log for every priority URL. Record the date, the affected query or intent, your diagnosis, and the substantive edits. That turns recovery work into a testable process. Without a log, several teams can modify the same page and leave you unable to connect later movement with a plausible cause.

Work in related batches rather than changing the entire site at once. Start with a small group that shares a documented weakness. Recheck query-level performance after those pages have accumulated enough impressions for a meaningful comparison. Keep the changes if the intended queries recover without harming conversions or accuracy; revise the hypothesis if they do not.

Do not judge success only by restored clicks. A revised page may attract fewer but better-matched visits. Review conversions, qualified leads, engaged visits, and the queries now associated with the page. The objective is durable visibility for the right need, not the recreation of every impression that existed before May 21.

Key takeaways
- Use May 21 through June 2 as the rollout window, and keep those dates out of your before-and-after comparison periods.
- Diagnose changes by page and query; total traffic alone cannot distinguish ranking, demand, click-through, and technical problems.
- Prioritize valuable pages with a clear loss and a specific weakness instead of rewriting the whole site.
- Improve intent match, distinct value, evidence, ownership, and internal linking before reaching for more schema.
- Measure Google rankings and AI-answer visibility separately, with a written change log for both.
Your next move is simple: export the two 14-day comparison periods, select the five affected URLs with the greatest business value, and write one testable diagnosis for each. Make only the changes that diagnosis supports. That gives you a recovery plan you can measure instead of a collection of update myths.

References
- CrushPress.AI – Google’s May 2026 Core Update Completes with Major Impact
June 2, 2026

SEO in the AI Era: What Changes and What Still Works

If you’re wondering whether AI makes your SEO program obsolete, the useful answer is no. It changes where discovery happens, how answers are assembled, and what success looks like. It doesn’t remove the need for accessible pages, clear information, credible evidence, or a recognizable brand.

Your job is expanding. You still need to help a page rank, but you also need to make its information easy for an answer engine to retrieve, interpret, trust, and represent accurately.

Key takeaways

SEO is evolving from ranking pages alone to making a brand and its knowledge retrievable across search and AI interfaces.
Technical access, search intent, useful content, internal links, and authority remain the foundation.
AI optimization adds clearer answer structure, stronger entity signals, supported claims, and structured data that matches visible content.
Clicks are no longer a complete scorecard. Track visibility, citations, brand representation, qualified visits, and conversions together.
Start with one commercially relevant topic cluster and improve the full path from question to evidence to action.

SEO has changed before, but the target is broader now

Early search optimization often focused on exploiting visible ranking signals. Practices such as keyword stuffing and cloaking could influence engines that were easier to manipulate. The landscape included names such as Excite, AltaVista, and Northern Light, and much of the discipline was learned through experimentation and informal community knowledge.

That model became less dependable as search systems improved. Panda and Penguin became major milestones because they forced site owners to confront content quality and manipulative promotion. The durable lesson wasn’t that optimization had stopped working. It was that tactics built around weaknesses in a system had a shorter life than work built around users.

AI is another shift in the interface, but it is not a clean break from search. A conventional results page gives a user several candidates to evaluate. A generative interface can combine information into a response before the user visits a website. Your page may influence that response, earn a citation, receive a click, or remain invisible even when it ranks well elsewhere.

This widens the optimization target. You are no longer working only for a blue-link position. You are working to become a reliable candidate whenever a system needs information about your topic, product, organization, or expertise.

What remains essential and what AI adds

A shared foundation connects organized web content on one side with AI retrieval and answer assembly on the other.

It helps to separate enduring SEO work from the additional demands of answer-driven discovery. If the foundation is weak, adding schema or rewriting a few headings won’t rescue it.

Area	Enduring SEO requirement	Additional AI-era requirement
Access	Pages must be crawlable, indexable, and internally connected.	Important facts must be available in readable page content rather than hidden behind an interaction.
Intent	A page should satisfy the reason behind a query.	It should also answer the follow-up questions a synthesized response is likely to combine.
Content	Information should be useful, original, and easy to navigate.	Definitions, distinctions, conditions, and conclusions should be explicit enough to extract without losing context.
Authority	Relevant links, reputation, and subject expertise support trust.	Consistent entity information and independent corroboration help systems identify who you are and why your claims matter.
Structured data	Valid markup can clarify page type and important attributes.	Connected, accurate entities can reduce ambiguity, but markup must agree with what a visitor can see.
Measurement	Rankings, impressions, clicks, engagement, and conversions show search performance.	Answer inclusion, citations, brand mentions, representation accuracy, and assisted discovery provide additional signals.

Do not treat the right-hand column as a replacement checklist. It is an extension of the left-hand column. A fast, well-linked, authoritative page with a precise answer is useful in either environment.

Build an AI-ready SEO workflow around real questions

A team organizes blank question cards, content modules, and source documents into a connected publishing workflow.

You don’t need to rebuild your entire site at once. Choose a topic connected to revenue, retention, or a recurring customer problem, then work through the following sequence.

Collect the language your audience uses. Pull questions from sales calls, support conversations, on-site search, keyword data, and Search Console. Group them by discovery, comparison, decision, and post-purchase intent. This prevents you from creating a disconnected page for every wording variation.
Choose one primary page for the topic. Decide which URL should carry the clearest, most complete answer. Merge overlapping material where it creates confusion, and use supporting pages only when a subtopic deserves separate treatment.
Put the answer before the expansion. State the central answer near the beginning. Then explain conditions, exceptions, evidence, examples, and next steps. A reader should not have to cross several promotional paragraphs to learn whether the page addresses the question.
Make important relationships explicit. Use consistent names for your company, products, services, people, and locations. Connect relevant author biographies, About information, policy pages, and supporting resources with descriptive internal links. Do not expect a machine to infer that two inconsistent labels refer to the same entity.
Add only defensible structured data. Select schema types that describe the visible page. Keep names, authorship, dates, offers, and organizational details aligned with the content. Validate the syntax, but also inspect whether the markup tells the truth. Technical validity does not correct a false or unsupported claim.
Strengthen the evidence layer. Replace vague assertions with demonstrations, documented methods, primary references, or clearly attributed expertise. Seek relevant third-party mentions because a claim repeated only across your own pages is not independent confirmation.
Design the next action. Match the call to action to the question’s stage. An educational query may need a related explainer or checklist. A comparison query may need specifications, constraints, or pricing context. A decision query may justify a demo, trial, purchase, or contact option.

Review the finished page as if its paragraphs might be separated from the layout. Check whether a definition still makes sense without the heading above it, whether a recommendation names its conditions, and whether a quoted fact remains connected to its evidence. This is good editing for people and useful preparation for machine retrieval.

Measure visibility without mistaking mentions for results

AI answers can change the relationship between visibility and traffic. A user may learn your name without clicking, or an assistant may cite your page while sending few visits. The opposite can also happen: a small amount of highly qualified traffic can produce meaningful business results.

Use a scorecard with four layers:

Search presence: impressions, relevant rankings, indexed URLs, click-through behavior, and the mix of branded and non-branded discovery.
AI presence: whether your brand appears for a stable set of important questions, whether it receives a citation, and whether the description is accurate.
On-site behavior: landing-page engagement, progression to another useful page, leads, sales, subscriptions, or other outcomes tied to the page’s purpose.
Business quality: lead relevance, conversion value, sales feedback, and the customer questions that remain unanswered.

Treat AI visibility checks as sampled observations, not permanent rankings. Responses can vary with phrasing and context. Keep a consistent set of questions, record the wording you used, and compare patterns over time. A single favorable response is not a strategy, and a citation that misrepresents your company is not a clean win.

Start with the strongest page in one valuable topic cluster. Clarify its answer, repair its evidence and entity signals, align its structured data, and give the reader a sensible next step. That work improves your odds across traditional search and emerging answer interfaces without betting your entire program on one platform.

References

Search Engine Land — Matt McGee on the wild west days of SEO

May 29, 2026

Tag: Authority

An automated audit finds evidence; it does not measure Google

Build a page-type-aware rubric before choosing a model

Run a browser-based evidence pipeline

Make every finding traceable and reviewable

Controls for predictable failure modes

Validate the auditor before expanding the crawl

Key takeaways

References

Map the operating footprint before choosing URLs

Build a hub-and-spoke system around customer decisions

The main locations hub helps people orient themselves

Regional hubs resolve a choice between facilities

Location pages represent real facilities

Service and location pages answer different questions

Make every proposed geographic page pass an evidence test

Consolidate geographic bloat without erasing useful local value

Add governance so the bloat does not return

Key takeaways

References

A useful shortlist is a map, not a podium

Key takeaways

Match each practitioner to the problem in front of you

SEO and GEO tied to lead generation

International and technical SEO

Quality, algorithms, and the search news cycle

A public expert and a delivery team are different purchases

Run this diligence before you hire an SEO or GEO expert

A five-part scorecard for the final decision

Promises that should stop the conversation

Build a roster that does not depend on one guru

References

Topical authority begins with clean page boundaries

Use Query Deserves a Page before creating a URL

Build the topical map around entities and query templates

Create local pages only when place changes the answer

Audit query ownership before expanding topical coverage

Key takeaways

References

The citation advantage is substantial but concentrated

A benchmark gives proprietary data a clear job

Retrievability turns a study into citation infrastructure

A bounded question and recognizable entities

A method readers can inspect

Limits, corrections and a stable home

Research planning should begin with the decision

Key takeaways

References

AI visibility begins with a legible entity

Authority combines consistency with differentiation

Co-citation reveals the authority network around a brand

Key takeaways

Build an evidence trail that systems can interpret

References

Authority now operates at three distinct layers

Choose the engine and prompt class before optimizing

Design passages around problems, claims, and constraints

Build off-site authority inside the relevant source network

Key takeaways

References

The conversion barrier is uncertainty, not simply price

Trust signals should answer the buyer’s next question

A trustworthy journey stays consistent from page to follow-up

Measure whether clarity changes customer behavior

Key takeaways

References

Anchor your diagnosis to the actual rollout

Build clean comparison periods

Separate ranking losses from other traffic losses

Find the losses that actually need intervention

Inspect the queries that disappeared

Look for patterns across winners and losers

Audit for satisfaction, not an imaginary update factor

Test whether the page fulfills its promise

Add value that another generic page cannot reproduce

Check ownership, duplication, and internal competition

Improve search and AI visibility without conflating them

Make controlled changes and preserve what you learn

Key takeaways

References