How we use Persona Insights to plan client content
How we find the buyer persona AI recommends a client to least, and turn that gap into a content plan.
A client can look healthy in AI answers and still be invisible to the person who signs the contract. The general view says visibility is fine. Then a sales lead mentions that procurement never seems to hear about the product. Both statements can be true, because AI answers change with who is asking.
This is the problem personas are meant to expose, and it is one of the more useful things we do for client content planning. Here is the process, starting with what you can do without any special tooling.
Why the average hides the gap
When you track a prompt set with no audience context, you get an answer written for a generic asker. A real buyer rarely asks generically. A finance lead frames a question around cost control and risk. A developer frames it around integration and documentation. A marketing manager frames it around reporting. The model shapes the answer to match, and so the brands it names can shift from one framing to the next.
An overall visibility number blends those together. If the brand is strong for one buyer and absent for another, the average lands in the middle and tells you nothing about either.
The manual method
Pick the two or three people who matter most in the client's sale. Write a short description of each: role, what they care about, and how they phrase a question. Then rewrite your most important prompts the way each persona would ask them, and run both versions.
Compare the answers. Which brands are named for each persona? Is the client named earlier, later, or not at all? What reasons does the answer give? The reasons are the content brief. If the answer for the finance persona recommends competitors "for predictable pricing" and never mentions the client, that tells you what the finance persona needs to see and where the client's content falls short.
Do this by hand for ten prompts and you will learn a lot. Do it for two hundred and you will stop doing it. That is where tooling earns its keep.
What changed in October
According to the Promptwatch changelog, a Persona insights page arrived on October 23, 2026. It shows how visibility, position, and sentiment change per persona. Each persona gets a radar chart set against the general audience, which the changelog defines as the answers that ran without a persona. The point is to see which audience AI recommends the brand to least.
Some details we find practical:
- The radar spokes can be topics or AI models, so you can ask whether a persona gap is about a subject area or about one engine.
- A card names the persona with the biggest gap and its weakest topic or model, which gives you a place to start instead of a chart to interpret.
- Personas are a new heatmap axis, and the persona radar is available as a custom dashboard widget and in PDF reports.
The last point matters for agency work, because it means the finding can go into the same scheduled report the client already receives.
How we turn the gap into a content plan
We work from the weakest topic for the weakest persona, and we keep the sequence short.
Read the answers behind the gap. The radar tells you where to look. The responses tell you why. We read a sample of the actual answers for that persona and topic and list the reasons the model gives for naming competitors.
Check what the client already has. Often the content exists but is written for a different reader. A product page aimed at developers will not help a procurement persona. We note which existing pages could be extended and which topics need a new page.
Write for the persona's question. A page that answers the exact question the persona asks, in the language the persona uses, stands a better chance of being retrieved for it. That means concrete detail such as pricing structure, security documentation, or implementation steps, depending on the persona. We do not stuff persona names into headings. The reader is a person, not a keyword.
Re-run and compare. After publication and crawling, we run the same persona prompts again and read the radar. We report the change in the persona gap alongside the shipped work, and we do not claim a page caused a movement unless the crawl and citation evidence supports it.
Where Promptwatch fits
Promptwatch is the platform we run client programs on, and Persona insights is the reason we can do this at the scale of a real prompt set. It handles the part that does not scale by hand: running the same prompts per persona, comparing against the general audience, and flagging the biggest gap. The persona setup itself, including who the personas are, is still our job and the client's, because a persona only works if it reflects the real buyer. We cover persona stacking and setup in more detail in our guide to creating prompt personas in Promptwatch.
The changelog does not name a plan for Persona insights, so we avoid promising it on a specific tier. Confirm access on the plan you are scoping before it goes into a client proposal.
What to say to the client
The client conversation is simple when you have the radar. We say: the general audience sees you at a certain level, this persona sees you at a lower level, the gap sits in this topic, and here are the three pages we propose. We give the plan a date for re-measurement.
We avoid a few overclaims. We do not say AI "prefers" a persona, because models do not have preferences, they have retrieval and training data. We do not guarantee that a new page closes the gap. And we do not use persona results as a forecast of sales, because the link from an AI answer to a signed contract has many steps outside our view.
A quick start for your own programs
- List the two or three personas who decide the sale, with the questions they actually ask.
- Compare the radar or your manual answers against the general audience, not against a competitor first.
- Start with the single biggest persona and topic gap.
- Brief content from the reasons the answers give, not from keyword volume.
- Measure again on the same prompts and report the gap, not only the average.
If the average looks fine and the pipeline looks odd, the persona view is usually the first place we look.