Ask as the customers you actually get

A persona is a constraint your customer is carrying. Check whether cost, hours or transport changes which businesses the engines recommend.

Two people can ask an assistant the same question and get different businesses back, because they are not really asking the same thing. "Who's a good dentist in Leeds?" and "Who's a good dentist in Leeds, I'm paying myself rather than through insurance?" are different questions, and you may be recommended for one and invisible on the other. Personas are how you find out.

What a persona is here

A persona is a constraint the customer is carrying, not a demographic.

It is not "busy professional, 35-44". It is "I'm on a tight budget", or "I don't drive", or "I need someone open at the weekend". If the funnel tells you where in their decision someone is, a persona tells you what they are dealing with when they ask. Both change the answer an engine gives.

This matters because it is checkable. Nobody can verify a demographic guess. Everybody can verify whether mentioning price changes who gets named.

Where you set your personas

Situations are set per location on its coverage page. Open Tracking > AI Engine Visibility and press Manage coverage, then choose the location (/dashboard/ai-visibility/setup), or open a location's Coverage tab and choose Edit coverage. In the app a persona is called a situation. The plan at the top of the page lists the situations it tests as chips: click one to see its wording and how often it has been asked, or the cross beside it to stop testing it. Press Add a situation to see the ones suggested for your business, the full list, and a box to write your own. Save with Save without running, or save and measure straight away.

Each situation adds one prompt to every need on the research map below the plan. Open a need to see the prompts your plan would ask for it before you save.

The plain ask is always on

The first row is Plain ask, and it cannot be switched off. That is deliberate.

The value of a persona is never its own score. It is the difference against the plain ask: recommended perfectly well when someone just asks, and invisible the moment they mention cost. Without the plain ask measured in the same batch there is nothing to compare against, and a portfolio with no baseline cannot produce a finding at all.

Important: Read persona results as a gap, not a grade. "60% visible" on a persona means little on its own. "80% on the plain ask, 20% once they mention price" is a finding you can act on.

For each location you can read that gap in three places: the Who is asking section on the Insights tab compares the plain ask with each situation over the same keywords, the keyword table on Summary says which way of asking a keyword loses on, and the Presence tab filters every response by who was asking.

Which personas you will be offered

SearchOps suggests the two or three constraints your own data supports, with the reason shown. On a first setup the ones your allowance can carry are already in the plan. If 31 of your reviews mention price, that is why you are being offered the cost persona rather than because it is on a generic list.

Everything else sits under More situations when you add one, grouped by what the constraint changes, so the question becomes "do any of these five apply to me?" rather than "pick from fifteen".

The suggestions fit the kind of business you are. An emergency callout trade gets "it's the middle of the night and I need someone now" and "I'm worried about the callout charge". A clinic gets "I'm paying for this myself rather than going through insurance" and "I can only get there after 6pm". A restaurant gets "there will be about ten of us" and "I'll have two young children with me".

You can also write your own, phrased in the first person the way a customer would say it.

Tip: Pick constraints you could actually do something about. If the cost persona shows you disappearing, that is a pricing-transparency job on your website and profile. If the out-of-hours one does, it may be that your hours are not published where an engine can read them.

What a persona costs

Each situation adds one prompt to every keyword the location measures, asked at the ready-to-buy stage only. One prompt put to one engine once is one AI visibility check, so one situation on a location with 20 keywords, measured on ChatGPT and Gemini, adds 40 checks to a run you start yourself. A scheduled run asks every prompt three times, so the same situation adds 120 checks to each scheduled run.

That is the real reason to be selective rather than tick everything. Two constraints that genuinely describe your customers tell you more than eight that might. Your run shows the total before anything is spent. See How credits and quotas work.

Next steps

Find the constraints your data already points at: Research your prompts.
Write the prompts themselves: Choose the prompts AI Engine Visibility runs.
Read the results: Track AI Engine Visibility.