Before I consider requesting a new SEO tool, I always ensure that I understand the trade-offs between custom solutions, SaaS platforms, and hybrid approaches that utilize both.
AI has empowered SEO teams, including mine, to become more ambitious about automation. Tasks that once required engineering support are now tackled easily with tools like Claude or ChatGPT.
This is thrilling, yet it brings a new challenge: the assumption that everything can be automated. In today’s language, it boils down to a single question: Do we build or buy the tool?
The build-versus-buy dilemma is intricate, made even more so by AI advancements. It isn’t merely about cost; it’s about security, maintenance, data access, internal capabilities, workflow fit, and whether a custom solution can stay reliable and useful as time progresses.
How AI Lowers the Barrier to Building
AI has drastically lowered the barrier to experimentation. Even those of us without technical know-how can now create custom GPTs, build workflows, connect data sources, or craft an internal AI assistant.
However, maintaining a tool over the years remains a challenge, even if I managed to build it initially with AI support.
AI significantly aids SEO teams in data analysis, pattern recognition, summarizing information, and recommending actions, saving us a lot of time. Ignoring AI would surely leave us trailing behind.
It’s essential to acknowledge that AI still hasn’t reached the level of human creativity. It excels at working from established patterns and predicting outputs. This could evolve in the coming years.
AI tools also come with unseen costs. Internally developed tools may appear free since their invoices typically bypass our SEO teams, but expenses from token usage, API calls, infrastructure, engineering time, security reviews, and maintenance do exist.
Many organizations, as noted by Reuters, are experiencing “AI sticker shock,” finding themselves unable to forecast usage-based AI costs accurately. Companies like Uber, reported by TechCrunch, have even established AI spending caps after exceeding their annual budget in only a few months.
Currently, marketing teams, including mine, aren’t the largest AI consumers compared to engineering teams. Yet, this could shift rapidly.
When this happens, our expenditures will undoubtedly rise, prompting organizations to evaluate which AI tools and processes genuinely add value as opposed to simply consuming our budget.
Start by Defining What You Need
Before choosing whether to build or buy, SEO teams must define their true needs.
Different Ways to Use AI and Automation
I’ve noticed that many teams, including ours, lump various solutions together, yet they differ in cost, complexity, and maintenance.
- A custom tool: Generally a complex internal system necessitating engineering support, often focusing on automation and potentially incorporating AI aspects.
- A custom workflow: A repeatable process built with numerous tools like a custom GPT, spreadsheets, and automation, usually with an AI layer.
- A custom layer on SaaS: Leveraging data from existing tools to shape personalized reporting, prioritization, or recommendation processes.
- A true AI agent: A system capable of taking more autonomous actions, such as scanning Slack and following up on pending communications.
Though similar, these are often misidentified. Overgeneralizing terms like “AI agent” can lead to cost and complexity misjudgments.
Look for Repetitive, Context-Rich Tasks
Our team is still exploring AI capabilities. So far, we have concentrated on daily tasks involving substantial manual work.
For instance, we developed a custom GPT to assess whether our content aligns with our personas and addresses their pain points. The aim is not to replace our copywriters or reviewers, but to ensure that content isn’t generic and suggest pertinent enhancements.
We’ve also leveraged AI for translations, monthly reporting, and creating a weekly summary that integrates meeting notes, Slack, and Jira to identify outstanding tasks or follow-ups.
One of our newest workflows converts internal meeting recordings into structured landing page briefs.
Such tasks are ideal candidates for AI-powered custom workflows, given their dependence on internal context, repeatability, and specific company knowledge.
Not Everything Should Be Built
A case from our team involved a colleague who vibe-coded a prompt tracking tool. Although a good start, data presentation required manual steps for trend graphing, soon becoming a maintenance hassle due to changes in LLM tools.
The core issue was reliability. For AI visibility and prompt tracking, we needed stable data presentation, leading us to switch to a specialized platform like Peec AI, rather than maintain our own version.
This experience was insightful, enhancing our understanding of the problem, complexities, and necessary features when considering external solutions.
Here’s my advice: whether opting to build or purchase a tool, always explore existing market solutions. It helps to narrow down the essential features, preventing reliance on non-essential ones.
Especially for business-critical tools like rank tracking and website crawling, smaller SEO teams without technical support should be cautious of building from scratch. Reliability should be prioritized when data is crucial for decision-making.
Use AI Where Your Data Already Lives
Consider buying a crawler, rank tracker, or AI visibility platform and focus on linking these with custom data like GA or GSC accounts, or CRM data. This integration allows comprehensive analysis in a single view.
MCP connections also warrant consideration. The Model Context Protocol is a standard for linking AI applications with external systems, enhancing current workflows.
Though not necessary to learn coding, understanding enough to ask the right questions is beneficial.
If sensitive data is involved, like proprietary research or customer details, it’s crucial to assess security risks. It may be safer to allocate engineer support to avoid compromising sensitive information.
Deciding on a custom tool requires acknowledging the full cost, including engineering time, security reviews, and API usage, despite invoices not being SEO-related.
Before requesting any tool, SEO teams should articulate the problem, expected value, cost comparison between building and buying, and potential consequences of taking no action.
Effective requests should not start with tool needs, but with the problem, its significance, tested solutions, and the proposed optimal solution.
How to Prioritize What to Build First
No one-size-fits-all matrix exists for prioritizing builds.
Tools vary; from website crawlers to content evaluation systems, each can’t be judged by identical criteria.
In doubt, start by mapping current workflows versus the ideal ones. Patterns often emerge, highlighting primary priorities.
The first group involves tools that aid revenue generation, like identifying content opportunities or improving conversion. Marketing, including SEO, seeks visibility and leads, thus revenue-centric tools can be higher priorities.
The second category concerns tools minimizing repetitive tasks. While they may not directly create revenue, they free up valuable team time for strategic work.
Quick wins should not be ignored. Stakeholders value timely results, thus a small project with potential returns within weeks can build trust and support larger initiatives.
Also, consider cross-team value in your decision. SEO problems often extend beyond one team. Collaborating with other teams can strengthen the business case for shared solutions.
Often, the best tool isn’t the most complex. Starting small could be the strategy for smarter progress.
Remember, effective scoping leads to good decisions. Even with AI easing the build process, proper scoping of what to build remains essential.
- Define the problem, expected value, user base, and post-launch maintenance.
- Engage with your team and other departments, identifying whether it’s solely an SEO issue or a broader business challenge.
- Avoid building for AI’s sake, or being swayed by impressive demos.
Neglecting scoping risks acquiring costly tools that don’t integrate with workflows or building internal tools beyond maintenance capabilities.
Thoughtful consideration of scope is crucial before opting to build, buy, or customize a solution.
Inspired by this post on Search Engine Land.
Leave a Reply