Our principles

What we will never promise you

No one outside the AI companies can see inside their assistants. We can’t either. So we don’t promise you rankings or growth. We measure how assistants answer questions from your field and suggest what to improve.

Check it yourself
Point by point

What we won’t promise, and what we do instead

  1. 1

    We won’t promise that AI will recommend you or that you’ll rank in Google.

    Why

    Their own systems decide who an assistant recommends and what Google shows at the top. No one on the outside can guarantee it.

    Instead

    We measure where you stand today. The score rests on full AI answers, and you can open every one of them.

  2. 2

    We won’t promise how much more often AI will mention you, or by when.

    Why

    We can’t control when an assistant comes across your change, or whether it uses it.

    Instead

    Every task shows our estimate of its impact and effort, and how confident we are in it. The task detail says plainly that it’s not a promise.

  3. 3

    We don’t prove success with one hand-picked answer.

    Why

    Ask AI assistants the same question and the answer almost always comes out a little different. A US study by SparkToro (January 2026) measured this for ChatGPT and Google’s AI. The chance that two answers to the same question gave the same list of brands was less than 1 in 100.

    Instead
    • We calculate the score from all answers in a measurement. Where in the answer AI named you doesn’t count toward it.
    • But a measurement with just a few questions is only a rough estimate.
    • In the app, a task based on AI answers is created only once there are enough of them. With fewer, we stay silent. That doesn’t mean everything is fine.
  4. 4

    We don’t take credit for your work.

    Why

    Between a change on your site and an AI answer sit many things we neither control nor measure.

    Instead

    When you mark a task that came from a measurement as done, we look at it in one of the later measurements. The task itself tells you whether you need to run that measurement yourself. A measurement can say only one of two things: “The finding is gone” or “The finding is still there”. Neither claims your change caused it. If a measurement can’t tell, we don’t claim anything.

What the measurement can and can’t do

Can

  • Show the full AI answers and the pages the assistant found or cited for them.
  • Show a plain count next to the score, such as “5 of 16 answers”.
  • Count mentions of the competitors you track in the same answers.
  • Show the tone in which AI wrote about you. The tone doesn’t count toward the score.

Can’t

  • Show exactly what a logged-in person sees in the ChatGPT app. We always ask with no history or personal settings.
  • Give the same result every time. An assistant may answer differently next time, even if you changed nothing on your site. A difference of a few points between measurements is often just noise.
  • Show sources for every answer. You only see them for answers where the assistant sent them to us.
  • Decide what a good score is. We set the band limits ourselves, based on what we see in measurements. They are not an industry standard.
  • Tell you how many visits or orders it will bring you.

Pricing, trial and cancelling

  • There is no annual plan and no annual commitment.
  • The trial lasts 7 days and you enter a card for it. After sign-up it runs on Profi. If you don’t cancel it within the 7 days, you start paying for the plan it runs on. Each account gets it only once.
  • You can cancel the subscription at any time. Under our terms, we don’t refund money for a period you’ve already paid for.
  • Measurements are paid with credits from your plan. Unused credits don’t carry over to the next billing period. If an assistant doesn’t answer, we refund the credits for that answer.
Public price list
a month, excl. VAT
StarterWe ask ChatGPT, Gemini, Google AI Mode and Perplexity.990 Kč
ProfiAdds Claude.1 490 Kč
EnterpriseAll five assistants.priced individually

Full price list

Check it yourself

  1. 1

    Open the full answer

    Every answer in the app has “Show the full answer”. You’ll see exactly what the assistant wrote. If it searched the internet, you’ll also see the pages it found or cited.

  2. 2

    Count the mentions

    The score is the percentage of answers in which AI named you: 5 mentions in 16 answers gives a score of 31.

    Questions asked directly about your brand count too, and AI usually names you more often there. That’s why the app also shows results by question group.

    One exception: for a question about your brand, we don’t count the mention when the answer shows that AI doesn’t know you. The app marks it on the answer.

  3. 3

    Read what each feature can’t do

    Every feature page has a “What this feature can’t do” section.

FAQ

Frequently asked questions

Why does this page exist?

A lot gets promised around AI visibility that nobody can deliver: sure rankings in AI answers, a guaranteed recommendation or quick growth. We want you to know exactly what you get from us, and what you don’t, before you start.

Is it worth improving anything if you guarantee nothing?

Yes. Assistants often list the pages they drew on for their answers, and that’s something you can work with. We show you what they write about you, who they recommend instead of you and which pages they cite, and turn that into specific tasks for your site. We can’t guarantee a result, but you won’t be improving blind: after a change, later measurements show whether the finding is gone.

I wrote this page as Custela’s founder. If you find a sentence on our site or in the app that contradicts it, write to me at info@custela.com.

Michael Cetl, founder of CustelaService provider, Company ID 08992207Updated

See your own numbers

Start 7-day free trial

Once per account, and you enter a card to start. If you cancel the subscription during the trial, you pay nothing. Otherwise you start paying for Profi, 1 490 Kč a month excl. VAT.