├─●─┤ citence

New — AI Overviews reached French-language queries on 24 July

What AI engines say about you, measured.

Perplexity and Google AI Overviews recommend brands to your customers. Citence samples those answers five times per question, every day — and publishes every figure with its uncertainty range.

Free audit · no account, no credit card · report sent to your inbox — or subscribe directly, 14-day trial.

app.citence.ai
Citence dashboard: overview of a tracked brand — citation rates by question family with confidence intervals, daily curve and a detected change point
brand awareness · sampled 25 July
95.9 %
[9398] · n=315

Good — cited in almost every answer.

21,488answers sampled
155,741citations parsed
96,774Google positions recorded
27 daysof measurement series

tracked set as of
28 July 2026

two markets, one method

In English it already happened. In French it started on 24 July.

AI Overviews present · English queriesEnglish-speaking market · 14-day window
86.6 % [84–89] · n=956
AI Overviews present · French queriesthrough 21 July 2026
0 % [0–0.1] · n=4 379
AI Overviews present · French queries25–26 July 2026
97.7 % [96–99] · n=389

We measure both markets with the same rules and the same ranges. In English, AI Overviews have been in place for months. On the French queries we track, they appeared on 24 July: one brand went from 0 to 97.7 % presence in 48 hours, and we have the exact date because the series was already running. Engines cite brands whose content they already know — whatever your market, your starting position is being set right now, and a measurement history cannot be reconstructed after the fact.

the product

Five screens, one question each.

The numbers are not decoration: they are the five chapters of your dashboard, in the order you open them.

  1. 01

    Overview

    Where I stand today, and what moved.

  2. 02

    Visibility

    Which questions cite me, and who replaces me.

  3. 03

    Trend

    Is it going up or down, and since exactly when.

  4. 04

    Action plan

    Why I lose, and what to do first.

  5. 05

    Ledger

    Every raw answer, readable, never rewritten.

app.citence.ai/plan
Citence action plan: diagnosis of lost questions with a measured cause, actions ranked by number of questions covered, badges marking targets that also rank in the Google top 10
The action plan, GEO + SEO. For every question where you are not cited, the cause is measured — authority, content, format, technical — then turned into ranked actions with their evidence. One single list: sources that also matter in the Google top 10 are marked.
app.citence.ai/visibility
Classic Google card: your site's positions in the Google top 10 per query, with the starting point of the series and the domains holding the ground
Your Google positions, from the same samples. Every sample of your questions also captures the Google results page — your SEO positions are measured on the same queries, retroactively from the start of your series, at no extra cost.

what makes us different

Measurements, not noise.

Honest statistics

Every question is asked at least 5 times per engine, per pass. Every rate ships with its confidence interval and its sample count. A tool that asks once is showing you a coin flip.

Your competitors, question by question

We do not just count your citations: we record who holds the ground in your place, question by question, along with the third-party sources that shape the answers.

A history you cannot buy back

The series starts the day you start. Raw answers are kept exactly as received and never rewritten: gaps in measurement are shown, never filled in.

how it works

Enter your domain. We measure. You read.

  1. Step 1

    Your domain, your email

    We read your site and build the 20 questions a future customer would ask AI engines — from "who are you?" to buying questions, with your competitors identified.

  2. Step 2

    The measurement runs

    Each question is asked 5 times to Perplexity and Google AI Overviews. Not a single draw: a question asked once can lie.

  3. Step 3

    Your snapshot, online

    You get a link to your report by email: where you are cited, question by question, who holds the ground, and the sources behind the answers. Free, no account, no credit card.

pricing

Three tiers. One price: yours.

The price follows your real volume — tracked questions, engines, measurement passes — with a published formula. Common configurations run from €29 to €549/month; on the pricing page, your exact amount is computed in front of you, and it is the amount checkout charges.

Startup — see

Your visibility measured, a confidence range on every figure, your classic Google positions included.

Pro — understand and act

The why behind every lost question, and a single GEO + SEO action plan, content briefs included.

Agency — operate

Several brands in one workspace, each with its own series, diagnosis and plan.

14-day trial · monthly, cancel anytime · annual −10%.

Build your price — the arithmetic is shown

frequently asked

What people ask before starting.

Why measure each question 5 times?

AI engines never give the same answer twice. A tool that asks once shows you a coin flip; we ask each question at least 5 times and show the rate with its confidence range. That is the difference between an anecdote and a measurement.

What is the range in brackets?

The confidence interval. "24% [21–27]" means the true value sits inside that range 19 times out of 20. The more samples accumulate, the tighter it gets — which is what lets you tell a real change from random variation. No figure ships without it.

Which engines do you cover?

Perplexity and Google AI Overviews today — plus classic Google: every sample also captures the results page, so your SEO positions are tracked on the same queries at no extra cost. More engines (ChatGPT first) arrive as integrations land — each new engine starts its own series, comparable to the others, and never raises the bill of a running subscription.

What is in the free audit?

A complete snapshot: 20 buyer questions generated from your site, each sampled 5 times per engine, with your rates by question family, the full list of questions (cited / not cited), who holds the ground, and the sources engines cite. Delivered as a web page by email — no account, no credit card, shareable with your team. A subscription then turns that snapshot into a film, with a 14-day free trial.

Where does our data go?

Measurements are taken on public engine answers — not on your internal data. What we collect (your domain, your email) stays with us: hosted in France, never resold, no marketing sequence behind the audit. Full detail in our privacy policy (French).

Can I cancel?

Whenever you want — the subscription is monthly with no commitment, managed online. Your series stay readable until the end of the paid period. History cannot be bought back: that is the only real reason to stay.

where to start

Get the free snapshot of your brand.

Your starting position, measured: the questions your buyers ask, who AI engines cite — you or your competitors — and where the open ground is. A web page sent by email, nothing to create or buy in order to read it.

You receive a link to your report by email. No account, no credit card, no marketing sequence behind it: your snapshot, then it is your call.