Persona Insights vs persona stacking: how they differ and when to use each
Persona stacking sets up the prompts. Persona Insights reads the results. A comparison of the two Promptwatch features, with a verdict and a who-should-pick-which guide.
Quick verdict
These two features are not rivals. Persona stacking is how you build a prompt panel that asks the same question as different buyers. Persona Insights, added on 23 October 2026 (changelog), is how you read what came back. If you only stack, you collect data you have to compare by hand. If you only look at insights without personas in your prompts, there is nothing to compare. Promptwatch is the tool we know of in our catalog that documents both halves, so this comparison is mostly about the order in which to use them.
The problem both features solve
You can be visible to a general audience and nearly invisible to the person who signs the contract. A broad prompt like "best accounting software" answers one way. The same prompt asked by a finance director at a 400-person firm, or by a freelancer with a tight budget, can name different brands. A single blended visibility number averages those differences away.
The fix is simple to describe. Ask the same questions as different buyers, then compare each buyer's results with the baseline. The work is in doing that consistently and reading the comparison without getting lost in spreadsheets.
Persona stacking: the input side
A persona is a short description of a buyer: role, company size, the job they want done, one constraint. Stacking layers several personas onto the same base prompt so you can see how the answer changes. Our guide to persona stacking and query fan-outs for prompt research explains how to write the personas and the seed prompts. The short version: three to five personas, written from real customer conversations, five questions each.
What stacking gives you is a richer prompt panel. What it does not give you is a verdict. After a week you have hundreds of responses split by persona, and the question "which persona is AI worst at recommending us to?" still needs an analysis step.
Persona Insights: the output side
According to the changelog, the new Persona insights page shows how your visibility, position and sentiment change per persona. Each persona gets a radar chart against the general audience, defined as the answers that ran without a persona. You choose topics or AI models as the spokes. A card points out the persona with the biggest gap and its weakest topic or model. Personas are also a new heatmap axis, and the persona radar is available as a custom dashboard widget and in PDF reports.
Reading a radar takes about a minute. If the persona's shape sits inside the baseline on one spoke, that is where you are weaker for that buyer. If it sits inside on every spoke, the persona is a bad fit for your current content, or your prompts for that persona are poorly written. Check the second possibility first.
Side by side
| Persona stacking | Persona Insights | |
|---|---|---|
| Role | Builds the prompt panel | Reads the results |
| Output | Responses per persona | Radar per persona against the general audience |
| Spokes or axes | Not applicable | Topics or AI models |
| Shows the biggest gap | No | Yes, a card names the persona and its weakest topic or model |
| Heatmap | Not applicable | Persona is a heatmap axis |
| Reporting | Underlying data | Dashboard widget and PDF section |
| Plan stated in the sources | Fact sheet lists it as a core feature | The changelog names no plan |
Sources: Promptwatch changelog, 23 October 2026, and the Promptwatch fact sheet we hold. The changelog does not state a plan for Persona Insights, so we name none.
How other tools approach personas
Our catalog notes that AthenaHQ lists persona targeting among its Enterprise additions, alongside an audit log and the Athena Citation Engine. It does not describe a per-persona comparison view, so we cannot say how that works. Other entries in our catalog do not record persona features. As ever, unrecorded means unknown. Our roundup of tools for buyer personas covers the wider field, and it is worth a read before you shortlist.
A workable sequence
- Write three personas. Include the one who signs, the one who uses, and one you suspect you are missing.
- Stack them on your ten most important prompts. Keep wording identical apart from the persona.
- Wait for a full period of data. A few days is not enough to see a pattern. A week or two is a sensible minimum for a first read.
- Open Persona insights. Start with the biggest gap card. Switch the spokes between topics and models, because a gap that shows up in one model only is a different problem from one that shows up in every topic.
- Put the persona on a heatmap axis. Compare persona against model or topic to find the exact cell.
- Decide on one action. Perhaps a page that speaks to that buyer, or a comparison for their situation. Then check again after you have published.
What to watch out for
Personas are only as good as the descriptions. A vague persona produces a vague answer. Equally, a persona that is too narrow produces answers from a tiny slice of the market. If the radar for a persona is far from the baseline everywhere, test the wording before you rewrite your site.
Also remember that the baseline is the answers that ran without a persona. That is not the "average buyer." It is how the engine answers when it knows nothing about the asker. Treat it as a reference line, not a target.
And do not over-read small differences. AI answers vary between runs, so a small gap on one spoke can vanish next week. Look for the persona with the largest and most persistent gap. That is what the biggest-gap card is for.
Who should pick which
- Stacking first if you have not defined personas yet. Without them there is nothing for the insights page to show.
- Persona Insights once you have personas running and need to see where the gap is, and to put it in a report for people who will not open the tool.
- Both, in that order if you sell to more than one type of buyer, which is most B2B teams.
- Neither if you have one clear buyer and a small prompt set. A plain visibility dashboard will do.
Bottom line
The decision is not which feature to pick. It is whether you need persona-level answers at all. If you do, start with stacking, then read the results with Persona Insights. Promptwatch is the one we would evaluate first, since it covers set-up and reading in one place, and puts the result into a dashboard widget and PDF that stakeholders can read.