How Peekl works
We check whether AI assistants recommend your business the way your customers actually ask, then hand you a ranked plan to fix what is holding you back. Here is the whole method, start to finish.
The four steps, in detail
We start from how people actually ask, not generic prompts. That means around 15 questions spanning nine buying intents, from “best [category] near me” to “a good [category] in [your area]” to “[category] for [a specific need]”. Each one is worded for your category and your city, because that is what your customers type.
Generic prompts give generic answers. Questions that match real demand are what reveal whether AI recommends you when it actually counts.
Every question goes to ChatGPT, Gemini and Claude, several times each, because AI answers change from one run to the next. That is 135 checks in a single audit; one screenshot tells you almost nothing.
So we measure how often you appear across all 135 checks, not a single lucky snapshot. That frequency is the difference between a hunch and a measurement.
When an assistant names a business, it is pulling from somewhere: a local list, a directory, a review thread. We capture the cited sources behind each answer, read your Google Business Profile and reviews, and read your own website the way AI does.
That is how you see why competitors get picked, where you are absent, and what is missing on your own site, so the fixes point at real places to be listed and real pages to add rather than guesses.
Naming you is only half of it. When an assistant does name you, we also judge the tone it uses about you specifically, positive, neutral or flagged, with a confidence on each read.
It is a secondary signal and a decision aid, not fact, so a cool or cautionary description becomes something you can act on rather than a number you stare at.
Two ways to lose to AI
They need completely different fixes, so the first thing your report tells you is which one is happening to you.
Customers ask, other businesses get named, and you are left out of the answer entirely.
You show up sometimes, but a competitor is the one it actually recommends.
What you get: a measurement, then a plan
You get a clear measurement of how often each assistant names you versus every competitor, and the sources behind those answers.
Then the part a score leaves out: a ranked plan of up to 12 fixes, quick wins first, each one made for your business. A fix names the competitor getting picked instead of you and the exact question you are losing, then gives you the do-it-yourself steps to close it. You also get a side-by-side of what your top competitors have that you do not. You are never left with a number and no idea what to do about it.
Your website, read the way AI reads it
We show what an assistant cannot see on your site: the customer questions you have no page for, and whether your name, address and hours are in code it can read. Each gap comes with the fix and, where it is code, the exact snippet to paste. Monthly tracking also flags what changed on your site since the last check.
How AI describes you
When an assistant names you, the audit reads the tone it uses, positive, neutral or flagged, with our confidence and a few real examples, and what to do if it is not flattering. A secondary signal and a decision aid, not fact. The audit is a snapshot; monthly tracking follows it over time.
Your Google signals
How complete your Google Business Profile is and how your reviews look. These drive Google's own AI directly and shape the directories and review sites the other assistants read, and weak reviews can leave a business out of local recommendations. Directional, not a guarantee.
Six Physio, London
ChatGPT + Gemini + Claude · 15 customer questions · run 3× each · 135 AI answers analysed
Even an established clinic gets named in only 19 of 135 answers, while its competitors get picked far more. Getting recommended by AI isn't automatic. It's something you have to work on, and this shows exactly where.
When an assistant names them, the tone is positive across 18 mentions. A secondary signal and a decision aid, not fact.
Strong reviews and a near-complete profile, yet still named in only 19 of 135 answers. Google's own AI leans on these, and the other assistants read the sites they shape. Directional, not a guarantee.
What we promise, and what we don't
We never promise a number-one spot, and you should be wary of anyone who does. AI answers change, and no tool controls them.
What we do promise is evidence and a clear next step. We show you what the recommended businesses have that you do not, help you close the gaps, and you can re-check your presence over time with optional monthly tracking.
More questions, answered
Do the AI answers even change?
Do you look at my own website too?
Do you show how AI describes me, not just whether it names me?
Do you check my Google Business Profile and reviews?
How long does the full audit take?
Do I need to give you access to my website or accounts?
Which AI assistants do you check?
What do I do once I have my fixes?
See where you stand
The free check takes a couple of minutes and needs your business name, city and website. It asks ChatGPT or Gemini, your pick, the questions your customers ask and shows whether it recommends you today, who it names instead, and how often. The $99 audit runs the same questions on all three assistants, Claude included.
See plans and pricing