Pricing
By quote, and here is why
A question set is built for a market and not for an industry: publishing a single price would mean publishing a wrong price for everyone. What drives that price, however, can be stated plainly — and can be seen, since the demonstration runs on your own brand, not on a sample dataset.
Two questions, two answers
Can an AI even read my site? That is the AI Score. Forty checks, a score out of 100, and under every line the exact request that justifies it: robots.txt crawler by crawler, a firewall refusing GPTBot while robots.txt allows it, sitemap, structured data, callable surfaces. The Benchmark sets it against your competitors, measured the same day, on the same checks. Ten minutes and you know whether you are ahead or behind, and on exactly what.
And what do they say about me? That is the campaign, and that is where the work begins. Your share of voice, the brands cited instead of you, the questions you lose, the factual accuracy of what is said about you. That answer exists nowhere: no tool holds it, no dashboard stores it. It has to be gone and fetched, one question at a time.
Concretely: your corpus is written for your market, in your buyers' own words — not translated from a neighbouring sector. Every question is put to every assistant, several times, because an AI never answers the same way twice and a rate without repetition is worthless. Every answer is read, brands recognised, claims checked against your own facts. Then the corpus is frozen, so that next month's measurement is comparable to this one.
That work is what you are buying, and it has no shortcut. What you get out of it, however, is simple: you stop guessing.
Figures that survive the meeting
Any indicator about AI visibility gets challenged within five minutes: "where does that number come from?", "what about the other brands?", "when was it measured?". Etalo is built for those five minutes.
Every finding carries its proof. Under each line of the score sits the request sent and the response received — reproducible with a single command. Under each citation rate sit the raw answers, kept as they came and exportable.
Nothing is scored that was not measured. A check that did not complete leaves the scale instead of counting as zero, and the report states how many points the score actually rests on. When too much is missing, no score is published at all: a figure you cannot defend is no better than no figure.
Comparisons are made on equal terms. In the Benchmark, the ranking covers only the checks measured across every domain — otherwise the most closed site would also be the best rated, and the ranking would reward opacity.
What is not charged for
The AI Score and the Benchmark, then, with no quote and no commitment. Very often they are enough: an Access axis at 10 out of 30 takes a few days to fix, and that is already half the subject settled. You read them inside the tool, through a read-only account created for your brand — the report stays at your teams' disposal, and it updates with every new measurement.
Additional accounts for your teams or your agency are not charged for. A read-only account takes a minute to create, per brand and per country. Charging per seat would encourage credential sharing, which makes the action log useless.
Exports are not charged for either. Everything exports to CSV at any time: the corpus, the raw answers, the indicators, the correction plan.
The four variables
A campaign's price depends on the volume of API calls it commits, and that volume can be calculated. These four parameters determine it entirely.
- Corpus size
- The number of questions representative of your market. The sector packs shipped with the tool hold 200, spread across the four stages of the buying journey: discovery, exploration, comparison, selection. A narrow market is well served by that; a broad range needs more.
- Assistants queried
- Each assistant counts as one or two pairs, depending on whether it answers from memory, with web search, or both. Querying ChatGPT, Claude, Perplexity, Google AI Overviews, Google AI Mode and Bing Copilot does not cost the same as querying two.
- Markets and languages
- A brand present in France, the UK, Germany and Italy means four distinct corpora — buyers' vocabulary does not translate word for word. Four markets cost four times one, with no artificial volume discount.
- Frequency
- A one-off campaign, or monthly tracking. Tracking only becomes interesting from the second campaign on: it is the gap between two measurements on the same frozen corpus that says something, not the first snapshot.
The arithmetic, to place the order of magnitude
A campaign's call count is the product of three numbers: questions × engine/mode pairs × passes.
A 200-question corpus, queried across 7 pairs, with 3 passes per question, therefore means 4,200 calls. Each question is asked several times because an AI never answers the same way twice: without repetition, no rate can be published with a confidence interval.
That volume is the main cost driver, and it is the same for everyone at equal parameters. This is why a quote takes minutes to build once the scope is known.
Two ways of working
| First campaign | Monthly tracking | |
|---|---|---|
| What you get | A complete snapshot: share of voice, competition by assistant, lost questions, derived queries, factual accuracy, correction plan. | The same, plus the gap with the previous campaign on an identical corpus — therefore a trajectory. |
| The corpus | Built and validated with you, then frozen. | The same one, month after month. That is what makes campaigns comparable. |
| When it helps | Once the AI Score has been read and the technical obstacles cleared: to find out what is said about you, and against whom. | To measure the effect of what you fix, and spot a competitor on the rise. |
| Commitment | None. A one-off engagement. | Cancellable; notice period stated in the quote. |
In both cases the price is fixed in the quote and does not change during the contract. API costs are included: you have no account to open with OpenAI, Anthropic or Google.
How we get to a quote
You describe your scope
Your ranges, your markets, your real competitors. By form, or in thirty minutes on the phone.
We compute the volume
Questions, assistants, markets, frequency: the four variables above give a number of calls, therefore a cost.
You receive a priced quote
Within two working days. It details the scope retained and what it produces, so you can compare it with something else.
The AI Score first
It reads in ten minutes. If it shows that AI crawlers cannot even get into your site, start there: measuring what is said about you can wait until you are readable.
Let us look at the tool on your brand.
A demonstration on your own site and your own competitors, not on a sample set. You see what an agent can read of your site and where you stand; the campaign comes afterwards, if it is warranted.