Many SEO teams are already using AI to write content. Almost none of them can explain the plan behind it.
In a recent SEJ webinar, Darrell Tyler, Senior Manager of Organic Growth at CallRail, shared statistics from his conversations across the industry: about 85% of SEOs are talking about using AI in content, and about 12% have written programs to manage that use.
That gap is the whole problem. The discovery has already taken place. What separates teams now is whether the AI is running on base or lagging.
Darrell walks through the four layers that turn AI subscriptions into real profit, why your content reads generic without them, and research that shows where your gaps are.
Watch the on-demand webinar right now and get the full outline.
85% of SEOs use AI in Content. 12% Have a Plan Behind It.
Discovery resolved. In Darrell’s discussions across the industry, most SEOs are already using AI in content in some way. The breakdown shows one layer down: only about 12% have written plans for how AI is actually used.
“If your use of AI is the same as your competitor’s use of AI, you don’t really have a strategy or an advantage, you just subscribe,” Darrell said.
The symptoms of a botched surgery are ones that most doctors recognize. The result comes out among the members of the group because everyone conducts their own information. Deterioration in quality in measure: the first few articles look good, then with article 97 there is a noticeable drop because the work started to prepare saved tokens instead of business results. Publish 500 articles on a weak basis and generate 500 poorly positioned pages, not 500 wins.
Darrell named this limited friction, the invisible quality of atrophy, and improved drift. Scaling AI without systems to support it is not growth. It costs real traffic and real time spent to repurpose published work.
The first move is an honest assessment of where your team really stands. Run AI growth tests within an on-demand session.
Why Your AI Content Reads Like Everyone Else’s
Why does AI content sound familiar?
Because the AI is starting from the same blank slate that your competitors are using. If you’re writing an article about what phone tracking is, and your competitor is writing the same article about the same information, you’re both sending almost the same output. Darrell calls the input “blank slate AI,” and it’s a big part of why AI content is being beaten from an organic perspective. It is the same as everything else that has been published.
The line he wants you to leave with: “You can’t figure your way out of an unwritten context.”
Rapid engineering is real, but it doesn’t redeem AI that has no context about your business. A model is not a bottle. The platform is not a bottleneck. Working around AI is something. Without the written context, the AI writes to what is available online, which is the same source as your competitors.
Action item: before you rate, write the context that makes your content unique: your product and brand positioning, your first-person data, and the angles only your team can provide.
Learn how written content looks in action, on the web of choice.
Teach AI Your Business Before You Ask It To Write It
What is AI Ops for SEO?
It’s a system that governs how AI produces consistent, high-quality, product-aligned work at scale. Darrell’s framework has four layers, which borrow heavily from MLOps and RevOps and refer to content.
The knowledge layer is the true source of your AI for your business: product and product ontologies, style guides, competitive intelligence, and first-person data such as reviews, customer stories, and call logs. He calls this the most important layer, because it is the one that adjusts the consistency of AI. The AI stops typing from the topic on its own and starts typing from your location.
The workflow layer is where the human power becomes the organizational level: SOPs, agile libraries managed as production code, templates. The governance layer is the people side: QA frameworks, review checkpoints, and feedback loops that build trust in outputs over time. The application layer, the tools and the models themselves, measure what is important. Models are the engines that you swap out when sailing. Your system does not change when the engine changes.
The first company’s data is the part that many groups skip and the part that gains the edge. Reviews, customer stories, and call transcripts give AI the first experience to write about, which are the very rewards of organic search.
The content of each layer, what must be included in the knowledge base, how to organize workflow SOPs, and how control checkpoints are removed as trust is built, are transferred to all necessary. See what goes inside each layer.
Stop Measuring Content by Volume. Start Measuring Results.
How should you measure AI content if not by volume? For the results it drives. A competitor can buy the same AI subscription tomorrow. They can’t buy the knowledge layer, workflow, and governance you’ve built and insisted on for a year. That’s the connecting part.
Darrell’s advice on tools is to remain LLM-agnostic by design. Start today’s work with any high-performance model, and when the leader changes, change the engine, not the performance. Store your assets, style guides, instant libraries, and deployment documents, living independently in a version-controlled environment rather than being locked within a single platform.
The role is changing too. Less writing from scratch, less manual checking, more strategy, building a knowledge base, and management. The technician becomes the system architect.
And the scorecard changes. SEO ROI is measured by efficiency, conversions, and revenue, not by how many articles you push out the door.
Watch the on-demand webinar for the full release, from research to implementation.
Q&A: Most Helpful Questions from the Webinar
Q: I feed AI links from my site. Is that enough to build a knowledge base?
Darrell replied: The beginning, not the end. Scraped links cover what is already public, but the value of the information layer resides in what is not on your website. He outlined the internal context such as the product portfolio, the audience you are trying to attract, and the positioning that cannot be done on the social page. Feed links, then dig deep into context AI can’t find on its own.
Q: Success information on ChatGPT is not a priority for Claude. How do I handle that?
Darrell replied: Information is only part of the good output. The other part is the unique context. If you have a strong sense of what looks good, lean on that and ask the AI to help you bridge the gap. He pointed out that if you provide the same unique context, you get a balanced result regardless of which model you use, making the instant difference across platforms irrelevant.
Q: Beyond impressions and clicks in the search console, how can I tell if my AI content is doing more harm than good?
Darrell replied: Go to GA4 to find the page and read the engagement signs. Average engagement time and views per user tell you how well the content performs once someone has arrived, not just whether Google ranked it. His informal litmus test: ask someone outside of work to read it, and if they struggle, the content probably isn’t strong enough.
Q: It’s been a year and my AI content is still average. Is it a command or a model?
Darrell replied: It’s not a model. Start with the command, then look more closely at how much context you’ve given the AI to do the job. His analogy: ask two people to build a house, and the one who asks if he wants brick or wood, who collects the core first, brings the idea to life. He who runs away quickly does not do that. Check the information, but check the context behind it, because the combination is what drives the output.
Watch the Full Webinar
Watch the on-demand webinar now.
