- Measure
- Fix
- Re-measure
Spot what AI gets wrong about your clients.
Prove the fix.
Every other day, RadarIA asks the AIs the questions your clients’ buyers ask — every day with the option. Every Monday, a report under your name: citations, errors spotted, proof of correction.
The AIs we query at every measurement
- ChatGPT
- Claude
- Gemini
- Perplexity
- DeepSeek
- Mistral
- Google AI Overviews
- Google AI Mode
An AI does not answer with ten links.
It cites a few names.
When a buyer asks for the best provider, the AI names a few brands. Is your client one of them? No SEO report tells you.
A recommendation made in a conversation leaves no trace on the site: only a possible click shows up.
ChatGPT, Claude, Gemini, Perplexity, DeepSeek, Mistral, Google AI Overviews and Google AI Mode, asked every other day with the questions your clients’ buyers ask.
The questions are asked again every other day, every day with the option. On Monday: a dated report, under your brand, per client.
Being cited is good.
Being cited correctly is what you can invoice.
SEO suites measure whether your clients are cited. None, to our knowledge, says why they are not, what the AIs say that is inaccurate, or proves that a correction worked.
1 · Access
Can the AI crawlers read the site? Blocking OpenAI’s two bots removes the site from ChatGPT’s answers. Across 3,054 small-business sites in 12 countries, it is rare in France (1.1%) but total; what is massive: 71.2% with no price, 74.9% with no update date. The Free diagnostic checks robots.txt (12 AI crawlers), headers and tags; it reports JSON-LD, llms.txt and noai without recommending them. No account needed, as a PDF.
2 · The fact sheet
Your agency writes what is true about the client, with sources. Without a fact sheet, no error is reported: we do not guess.
3 · The error found, then the proof
Each answer is checked against the fact sheet; an error only counts with its excerpt, and you confirm it. Once the correction is published, the question is asked again after 30 days to the 8 AIs: the error is gone or not, and the report says so.
Promise an error will disappear, or publish without your approval and the brand’s signed authorisation.
Measured in French, English, Spanish and Arabic
AI tracking tools mostly measure English. RadarIA asks its questions in French, English, Spanish and Arabic, every other day: it is your brand’s watch in the AIs, so that weak signals — a brand no longer cited, a competitor taking its place — show up on Monday.
From adding a client to the Monday report: three steps.
Add a client
Name, competitors, sector, country. RadarIA suggests 15 buyer questions; you adjust them.
The 8 AIs are queried every other day
Each question is put to ChatGPT, Claude, Gemini, Perplexity, DeepSeek, Mistral, Google AI Overviews and Google AI Mode: who is cited, in which position, in what tone, with which sources.
The report arrives every Monday
Score, competitors, lost questions, sources: under your name and logo, sent to you or to your client.
What the report also carries
The sources the AIs cite for your client: where the agency can act.
The tone of each mention: positive, neutral or negative.
And on Google? Search Console linked (read-only) or CSV export: clicks and impressions next to the AI share of voice.
Every published page is tracked at D+7 and D+30. A missing measurement reads “pending”, never zero.
Three benefits, included in every plan
A measurement close to real use, alerts that don’t cry wolf, and a way to find your next clients. Every limit is written next to it.
A measurement close to real use
ChatGPT, Claude and Gemini search the web before answering, as in their apps. Each client has its country and language; the report says, AI by AI, whether the country was actually passed on. Visits coming from AIs are read in Google Analytics 4. This is a minimum: some AI assistants don’t say where the visitor came from.
Alerts that don’t cry wolf
New competitor, lost source, recurring error, score drop: no alert on fewer than 3 measurements, within the reliability margin or because an AI was down. By email or signed webhook (Slack, Teams…).
A way to find your next clients
A prospecting document with your logo and no mention of RadarIA: a quick measurement on the typical questions of a prospect’s sector and a read of their site. And public pages per brand, with a badge to put on its site. A brand that is an agency’s client only gets a page with its consent.
Gemini receives the country in the wording of the question, Mistral and DeepSeek don’t receive it, and DeepSeek answers without web search; Google Analytics only counts visits whose source AI is known; the prospecting document is a one-off measurement, not tracking.
From the lost question
to the page online
For each brand where you turn it on, RadarIA writes the missing answer, publishes it in the brand’s name and measures the effect. Other brands stay measured.
1 · The draft
One page per lost question — answer, FAQ or comparison — written from the measurements and the site’s facts. Up to 3 per brand per week.
2 · The review
Unsourced figures and “[to be completed]” are flagged. You approve — or autopilot publishes after 48 h without a refusal.
3 · Publishing
Always after the authorisation the brand signs online, once. Where the page goes depends on the client’s site: see below.
4 · Measuring again
Measured again at D+7 and D+30, shown in Monday’s report. One-click removal. Nothing is promised.
Where the page is published
WordPress, Webflow, Shopify
RadarIA publishes itself, with content-only access the client gives you: WordPress application password, Webflow token, Shopify app. Revocable at any time.
Custom site, in-house tech team
RadarIA never touches the code or the repository (GitHub). You hand the ready page to the client’s team; once it is live, you paste its address into RadarIA and the re-measurement follows.
No access to the site at all
An answer centre in the brand’s name (answers.brand.com): a single DNS line to add.
Turned on or off brand by brand, prorated for the month. From the Agency plan.
Example: Agency plan (10 brands) with the option on 3 of them = €549 + 3 × €79 = €786 excl. VAT/month.
No testimonials yet.
Verifiable proof instead.
The deliverable, before you pay
The exact report format, freely available — demo data labelled as such.
The method, published
How the score is calculated, what it says and what it does not: all published.
Guaranteed, in writing
If your first Monday Radar teaches you nothing, the first month is refunded: request by email within 14 days, once per customer (Terms, article 6). Cancel at any time.
Can AI bots read your site? The Free diagnostic checks it.
Enter a site’s address: RadarIA reads its robots.txt and homepage, then tells you whether anything stops AI bots from reading it. No account, no credit card.
What it checks and what you receive
No account, no card
No sign-up and no payment method: only the site’s address. No call to an AI model; the diagnostic can be re-run after every fix, up to 5 a day per IP address.
What it checks
AI bots’ access: the robots.txt bot by bot (the ones that answer, the ones that train), noindex and nosnippet, content that depends on JavaScript, a written price and a visible update date.
The verdict at once, the details by email
The verdict (blocking, to check, or open) and the points found appear straight away. The details of each point and the bot-by-bot status arrive in exchange for your email address: no password, an unsubscribe link in every message; if the diagnostic finds an obstacle, up to two follow-up messages.
It does not measure whether AIs cite a brand, and it does not promise that a fix will get the site cited — it removes the verifiable obstacles. Asking AIs the buyers’ questions on a regular schedule is RadarIA’s monitoring, which starts with a subscription.
The price follows the number of clients tracked.
From €79 excl. VAT/month for one brand.
Every plan includes the 8 AIs, reports in French, English, Spanish and Arabic, a measurement every other day (daily as an option) and the pieces to publish. Essential and Solo tell you what to do; from Agency up, the Sunday Writer prepares it for you — and, with the Correction option, RadarIA publishes it.
Free trial: create your agency account and add a brand — RadarIA asks its questions once to ChatGPT, Claude, Gemini and Perplexity, no card needed. One trial per agency.
Essential
Solo
Agency
Network
Group
Correction option
Measure, fix, measure again: RadarIA publishes the approved pieces on your clients’ websites — or on a subdomain in their name, with no access to their site — always with their signed authorisation. Automatic review, optional autopilot, one-click withdrawal. No result promised.
From the Agency plan · brand by brand · no commitment
- All 8 AIs: ChatGPT, Claude, Gemini, Perplexity, DeepSeek, Mistral, Google AI Overviews and Google AI Mode
- 15 editable questions per brand, asked every other day to the 8 AIs
- Monday Radar: the report under your brand, every Monday
- Scores, trends and competitor citation share
- Pieces to publish: a brief and a first draft for every lost question
- Reports in French, English, Spanish or Arabic
- And on Google?: the client’s Search Console (read-only connection or CSV import) next to their AI share of voice
- Tracking of every published piece: before, D+7, D+30
- And from AIs?: visits coming from AIs read in Google Analytics 4 (read-only or CSV import), a minimum
- Alerts: new competitor, lost source, recurring error, score drop — by email or signed webhook
Pricing terms
- Every other day
- Each brand is measured every other day: that day, all its questions are asked to the 8 AIs. The week’s score rests on 3 or 4 measurement days, and its margin of error is shown next to it.
- Daily measurement option
- A brand measured every day instead of every other day: +€49 excl. VAT a month for that brand, pro rata. Turn it on and off brand by brand, from the brand’s page.
- Sunday Writer
- Every Sunday evening, RadarIA prepares for each brand up to 3 drafts on the questions where it is not cited, ready to review on Monday. On Solo, drafts are requested one by one.
- Localities
- A question can carry a city (“best plumber in Lyon”): useful for chains and networks. On Solo, questions stay national.
- 30-question option
- Every plan measures 15 questions per brand. With the option, a brand follows 30, measured on the same schedule (every other day, or every day with the Daily measurement option): +€49 excl. VAT a month for that brand, pro rata. Turn it on and off brand by brand, from the brand’s page.
- Correction option
- RadarIA publishes approved pieces on the client’s site, or on an answer centre in their name, with the brand’s signed authorisation, then measures them again. From Agency, brand by brand.
The questions agencies ask us
Exactly which AIs are monitored?
ChatGPT, Claude, Gemini, Perplexity, DeepSeek, Mistral, Google AI Overviews and Google AI Mode — the 8 AIs that consumers and professionals use to search for products and services. All 8 are included for every tracked brand, whatever the plan (1, 3, 10, 30 brands or more).
Are the reports really white-label?
Yes: your name and logo on every report, and your name as the sender. Only the sending address stays on radar-ia.io, which is what makes the report arrive. Your clients’ replies come to you.
How much can I charge my clients for this?
That’s your price, not ours. To set it, start from what RadarIA costs you per tracked client: €99 excl. VAT a month for one brand on Essential, €249 for 3 brands on Solo, so €83 per client (€66 at the €199 founder price for 12 months, while seats remain), and €60 per client on Network (€1,790 for 30 brands). The report carries your brand: whatever you charge above that pays for the work you build around it — content, reviews, product pages.
Can I pay by card or by bank transfer?
By card or SEPA Direct Debit, yes: payment goes through Stripe, on a secure payment page. With direct debit, you enter your IBAN once and the subscription is debited every month; your bank confirms the first debit within 3 to 5 business days. Prefer a bank transfer? It is possible on request: write to us at contact@radar-ia.io, we agree on the terms together and the invoice is sent to you before the start of each month.
Who publishes, you or us?
You, by default: RadarIA drafts, you publish. With the Correction option, RadarIA publishes itself on WordPress, Webflow or Shopify (content-only access) or on a subdomain in the brand’s name, with its signed authorisation. Custom site: RadarIA does not touch the code; the client’s tech team publishes the ready page, and the re-measurement follows.
Can I turn on correction for some clients only?
Yes, brand by brand, from the Pieces page: €79 excl. VAT a month per active brand, prorated, on top of your plan (from Agency). The others stay measured.
How often are brands scanned?
Every other day for each brand: that day, its questions are asked to the 8 AIs. Every day with the Daily measurement option (+€49 excl. VAT a month per brand). A summary under your brand every Monday morning.
How many questions per brand?
15 editable questions per brand, on every plan, asked to the 8 AIs at every measurement (every other day, or every day with the option). To follow 30 on a brand, turn on the 30-question option from that brand’s page: +€49 excl. VAT a month per brand, pro rata, on top of your plan. The 30 questions are measured on the same schedule; you can turn the option off whenever you like.
Does RadarIA show Google traffic too?
Yes, at no extra cost: link the client’s Search Console (read-only, re-read every night) or upload its CSV export. Impressions and clicks appear next to the AI share of voice. A limit written under every table: Search Console does not separate Google’s AI answers from classic results.
Do the AIs search the web during measurement?
Yes, those that offer it: ChatGPT, Claude and Gemini are queried with web search on, Perplexity and Mistral search by design, DeepSeek answers without it. The client’s country is declared to the search tool of ChatGPT, Claude and Perplexity when it accepts it, Gemini receives it in the wording, Mistral and DeepSeek don’t receive it: the report shows this AI by AI. Measurement keeps its schedule (every other day, or every day with the option), and every score keeps its margin.
Does RadarIA measure visits coming from AIs?
Yes, at no extra cost: link the client’s Google Analytics 4 read-only (read every night) or upload the CSV export of the “Traffic acquisition” report. You see visits arriving from ChatGPT, Perplexity, Gemini, Claude and other assistants, by assistant and by landing page. This is a minimum: some AI assistants don’t say where the visitor came from, so real traffic from AI is at least this figure, probably a little more.
Does a one-day drop trigger an alert?
No. No alert goes out on fewer than 3 measurements, when the change stays within the score’s reliability margin, or because an AI was down; a new competitor must appear in two measurements in a row, and a score drop is compared week on week, with the same AIs. You choose the alert types, the frequency (immediate, daily, weekly) and the channel: email or signed webhook (Slack, Teams…).
What is a public brand page?
A page hosted on RadarIA showing what AIs answer about a brand: score with its margin and date, brands cited alongside, excerpts from the answers, method and limits, plus a badge to put on its site. A brand that is an agency’s client has no page without its consent. Each page lets anyone report an error or request a new measurement; the score is never changed by hand.
Can I cancel anytime?
Yes, at any time: from your dashboard, the Stripe billing portal or by email; the service continues until the end of the paid period, with no refund for the month already started. If your first Monday Radar teaches you nothing, the first month is refunded on request by email within 14 days, once per customer (Terms, article 6).
Add a client today. Once you subscribe, their first report goes out the following Monday.
The report carries your name and logo. It shows the client’s score, the competitors cited in their place and the questions to work on.