Key Highlights
- AI systems do not start from keywords. They start from situations, and content built around real decision scenarios outperforms content built around keyword clusters.
- Prompt volume data does not exist. Any tool claiming to show high-volume prompts is showing you Google search volumes but hardly any ChatGPT usage data.
- Talking to one expert in the industry produces better, more prompt insights than any keyword tool currently on the market.
- AI systems trust content that acknowledges trade-offs and limitations more than content that pretends a single solution fits every situation.
- Decision content mirrors how buying actually happens. People do not choose software by reading rankings. They choose by eliminating risk.
Here is a YouTube video on search engineering masterclass day 7:
How does content creation actually change in AI search?
The traditional SEO content workflow has been the same for over a decade. Research your target keywords, identify the queries people are already searching for, or look at the People Also Ask (PAA) section for an existing search query, build an outline, and write to match the intent.
The output is generally consistent, the process is quite predictable, and the metric for success is ranking position. Sounds simple, right?
However, Day 7 in our search engineering masterclass series introduces a different approach. The first article for the Megalist Project was built using a content model that does not start from keywords at all.
Instead of asking which keyword to target, the question was what specific decision a real buyer is trying to make and what content would actually help them make it.
Here is how the two approaches separate when it comes to building a single piece of content.
The table below shows the structural difference between SEO content and decision content across each step of the writing process.
Both approaches still produce content. The structure of the content, the inputs that shaped it, and the outcomes it is built to achieve are completely different.
Why does keyword research not work the same way for AI content?
Keyword research depends on volume data. Search volumes for Google queries are reliable, well-tracked, and publicly available through dozens of tools. Prompt volumes for ChatGPT, Gemini, and Perplexity are not.
No tool currently on the market actually measures how many people asked a specific question to an AI system, which prompts are trending, or how often a particular phrasing was used.
Any tool claiming to show high-volume prompts is showing you Google search volumes dressed up in a different language. Optimising for those numbers and calling it AI optimisation is guessing. The volume of possible prompts an AI system could receive is functionally unlimited, and the variation in phrasing across users is far wider than the variation in Google searches for the same intent.
People claiming to optimise for high-volume prompts are simply guessing. Whatever prompt data they are using was not built to track AI usage in the first place.
What replaces keyword research when prompt volume does not exist?
Real conversations with people inside the actual industry. The replacement is unglamorous, unautomated, and surprisingly effective.
For the first Megalist Project article, the topic was CRM software for clinical research organisations. Rather than conducting keyword research, Senthil shared a list of potential questions with a friend who works at a CRO.
The question put to him was simple: Which of these prompts would you actually type into ChatGPT or Gemini when you are evaluating a CRM for your business?
The prompts he immediately confirmed were not keyword variations. They were structured decision questions:
- What CRM works best for a mid-size CRO?
- Which CRM do CROs usually use for business development?
- What CRM is suitable for a CRO dealing with sponsors and biotech clients?
None of these would surface as high-volume keywords in any SEO tool. All of them are real prompts that real buyers actually use when making real decisions.
Five minutes of conversation with one expert in the industry yielded better prompt insight than any keyword research tool would have for the same topic.
How is decision content structured differently from keyword-led content?
Decision content is structured around scenarios, not clusters. Every section directly answers a real question a buyer is likely to ask at a specific point in their decision process.
The first Megalist article does not start with a list of the best CRM tools. The opening section establishes context. Buyers in clinical research organisations face long sales cycles, relationship-led business development, and multiple stakeholders inside every deal.
Setting up that context before introducing any solution is very important because AI systems do not just retrieve content. They evaluate whether the explanation fits the situation at hand.
Scrolling through the article reveals a second deliberate choice. There is no ranking of tools. There are no claims about which CRM is best. The structure walks through trade-offs. When a particular type of CRM works well, when it breaks, what kinds of buyers it suits, and what kinds of buyers should look elsewhere. Honesty about limitations is the design principle.
AI systems trust content that acknowledges trade-offs more than content that pretends a single solution fits every situation. The reasoning is straightforward.
A model evaluating sources for inclusion looks for explanations it can use confidently, and an explanation that openly addresses what does not work for whom is more reliable than one that paints every option as universally suitable.
Why should you focus on buyer content that matches the situation that happens during buying?
People do not choose software by reading rankings. People choose by eliminating risk.
Buyers in any considered purchase, especially in B2B environments with long sales cycles, work through their decision by ruling out options that clearly do not fit their situation. The buyer evaluating a CRM for a 200-person clinical research organisation is not looking for a list of the top 10 CRMs.
They are looking for content that helps them quickly identify which options can be eliminated, which deserve a deeper look, and which trade-offs they will be living with regardless of the choice they make.
Content built around this elimination logic mirrors the actual cognitive process buyers use. Coincidentally, the same logic is what AI systems use when they construct an answer.
The model is also eliminating sources that do not fit the situation, retaining sources that clearly explain trade-offs, and building the response from explanations that genuinely reduce uncertainty.
Decision content fits both the human buyer and the AI system because it is built around how decisions actually get made, not how content is conventionally structured.
What is the shift from SEO content to search engineering content?
Search engineering content is built backwards from the decision rather than forwards from the keyword. The starting point is the buyer scenario. The middle is the trade-off map. The end is a clear, honest assessment of which option fits which situation.
A full disclosure matters here. The first Megalist Project article is an experiment. There is no official playbook from any AI system specifying exactly how to structure content for inclusion. Whatever conclusions emerge from this experiment will come from observation, not from an established standard operating procedure published by anyone.
Treating this as proven methodology would be misleading. Treating it as a working hypothesis worth testing publicly is honest.
What is the shift from SEO content to search engineering content?
SEO content is built around a keyword. Search engineering content is built backwards from a decision a real buyer is actually trying to make.
The CRO and CRM example in this episode clearly shows the difference. A keyword-led approach would have produced a page titled best CRM for clinical research organisations, optimised it for ranking, and structured the content around informational intent. A decision-led approach starts with a CRO leader at their desk, weighing trade-offs among vendors that handle sponsor relationships differently, and writing content that helps the leader rule out the wrong options faster.
Both pieces of content might cover the same vendors. Only one of them is used by AI systems to construct an answer for that buyer, because only one is structured around how the decision is actually made.
This is the core of what search engineering is built to do as a discipline. Keyword research, technical SEO, and authority signals continue to do their job at the retrieval layer.
Search engineering operates above that, ensuring the content retrieved is structured to fit the reasoning paths AI systems are building, rather than just the ranking criteria search engines apply.
Do you want more traffic?
Why Does ChatGPT Sometimes Use Your Content Without Mentioning Your Brand?
.jpg)
Why Do Strong Brands Suddenly Vanish From AI Answers? Search Engineering Masterclass, Day 6
.jpg)