Put your brand inside AI assistant answers.
Georadar queries ChatGPT, Perplexity, Gemini and Claude on the questions your customers actually ask, sums up your position in a score from 0 to 100, then hands you the lever: 70 editorial sites we run ourselves, where your brand enters the articles assistants read.
A composite score from 0 to 100 aggregating presence, recommendation, share of voice, primacy, stability and tone. The grey band marks the period before any content was published.
Share of voice divides your mentions by all mentions of every brand cited. Across twelve weeks, the gap to the leader narrows from 22 points to 1.
Each row is a prompt actually run across the four assistants. The gap shows how many share-of-voice points there are to make up.
Georadar records every brand cited, not only those you declared. One click promotes a detected brand to a tracked competitor.
The shift under way
Six searches in ten now end without a click
Your customer no longer scans a list of links: they ask a question and receive a written answer naming three or four brands, so the visibility contest now plays out inside a text you never see and that nobody records on your behalf.
You are not being cited
Your competitors turn up in ChatGPT and Perplexity answers on your customers' questions, while nothing in your current dashboards flags it.
SEO alone no longer covers it
Assistants write their own summary from several sources, so ranking first on Google no longer converts automatically into a recommendation.
You are working without a reference point
Without a regular reading, you cannot tell which prompts you appear on, which ones competitors hold, or which way the picture is moving month to month.
How answers get built
An assistant answer is built somewhere other than your site
When an assistant compares brands, it leans on editorial content published outside the advertiser, since your own site carries authority on who you are rather than on how your category ranks. That is exactly where the opportunity sits, provided you hold a relevant editorial inventory and a way of measuring what it produces.
The usual approaches
- Optimise your own site, then wait
- Pitch publishers one by one, with no control over the publication calendar
- Publish across cashback or couponing inventory, which assistants do not consult when recommending a brand
- Measure without being able to act, or act without being able to measure
Georadar
- 70 thematic editorial domains, each with an audience of its own
- Published and hosted in-house, so publication is decided rather than negotiated
- Advice-led content, the format assistants pick up most readily
- Measure, publish, measure again, inside a single tool
Our proprietary score
Your generative visibility fits into one figure
The Radar Index aggregates six axes recorded separately on every prompt and every engine, with published weightings, so you can follow a trajectory without ever losing the detail that explains the movement. The result is a manageable indicator, comparable week to week and solid enough to present to a board.
Presence
Do you appear in the answer at all?
Recommendation
Are you advised, or merely listed?
Share of voice
What room do you hold against other brands?
Primacy
How early in the answer do you arrive?
Stability
Is your presence consistent?
Tone
In what terms are you described?
Features
The whole chain sits in one tool
The whole chain sits in one tool, from the prompt-by-prompt reading through to content going live on the network, which saves you assembling a monitoring tool, an editorial agency and a media house. The whole chain stays under one team's control, which makes the setup markedly more efficient to steer.
Share of voice per prompt
Each customer question is run several times on each engine, so the variability of the answers is smoothed out before anything is counted.
Cited or recommended
Appearing in a list and being actively advised are two commercially very different situations, counted here by two separate rates.
Tone of the answers
Every mention is graded positive, neutral or negative, so you know in what terms you are discussed and not merely whether you are discussed at all.
Sources cited by the engines
A ranking of the domains assistants lean on to answer your prompts, with your own domains flagged. It is the map of the ground to cover.
Competitor detection
Every brand cited is extracted, including those you were not watching, and promotes to a tracked competitor in one click with its history intact.
Rank and stability
A citation opening an answer is not worth the same as one closing a list, and steady presence beats erratic presence. Both are recorded.
Analysis explorer
The movement of your Radar Index, your share of voice and your rank can be viewed by prompt, by category or by engine across the whole tracked period.
Prompts drawn from your queries
The Search Console connector turns your real queries into conversational questions, which analysis of your site then completes. Your measurement plan stops being guesswork.
Writing and publishing
Articles and inserts are produced from the platform, aimed at the prompts where you are missing, and go live on the most relevant site in the network.
AI traffic pixel
The declared assistant crawlers (GPTBot, PerplexityBot, ClaudeBot, Google-Extended) are separated from human traffic across all your published pages.
Baseline measurement
When a campaign opens, the current reading is frozen, so every subsequent movement is read against the same starting point.
Multiple teams and brands
Your colleagues and agencies join with distinct roles, and several brands coexist in the same workspace, each with its own prompt plan.
The assistants queried at every measurement
Connectors
Your Search Console queries become your measurement plan
Georadar plugs into the tools you already run, so the prompts tracked come from your real queries and the sessions referred by assistants sit alongside your conversions. Your measurement plan gains in relevance, since it starts from actual demand rather than from a list of imagined questions.
Google Search Console
ConnectorYour real queries become your measurement plan. Georadar imports your Search Console queries, groups them by intent and derives the conversational questions your customers put to assistants.
- Queries and their volumes imported
- Derived into commercial prompts
- Prioritised by volume and intent
Google Analytics 4
ConnectorThe loop closes on your own figures. Sessions referred by assistants are matched against your GA4 conversions, so generative visibility can be read alongside your other channels.
- Assistant-referred sessions isolated
- Matched against your conversions
- Compared with your other channels
Georadar pixel
ProprietaryOne line of code on your pages, and you watch the assistant crawlers go by. The pixel separates human traffic from declared crawlers and timestamps every visit.
- GPTBot, PerplexityBot, ClaudeBot, Google-Extended
- User-agent signatures maintained by us
- Deployed by default on network articles
Our inventory
We publish the 70 sites in the network ourselves
R-Advertising runs a network of thematic media established over many years, precisely the kind of source assistants consult at the discovery stage, provided the content there is editorial rather than promotional. It is a rare asset in this market, and it is what makes the offer genuinely executable.
25 editorial verticals covered
Activation
Two formats for entering the articles assistants consult
Publication happens in context, within the editorial line of the site hosting the content, because a disguised advert is spotted immediately and serves neither the reader nor the citation. Useful content remains the most effective route into the answers over the long run.
The dedicated article
A full piece written around a specific prompt where your brand is missing, published on whichever site in the network matches the question most closely.
- Editorial tone matched to the site
- Mention and recommendation brought in naturally
- Indexing tracked as crawls come through
- AI traffic measured by the built-in pixel
The in-context insert
A short editorial block placed inside an existing high-traffic article, as an expert viewpoint, a data panel or a piece of directly usable advice.
- Quick to publish on an established article
- Contextual to the page, credible to the reader
- Lower unit cost than a full article
- Suited to peak seasons and tactical campaigns
Method
Every publication is measured before and after
The mechanism runs in four repeating stages, and the weekly cadence keeps the curves comparable from one week to the next since the reading always happens at the same moment. That regularity is what makes the measurement reliable and the decisions solid.
Measure
Your strategic prompts are run across the four assistants, with several repetitions, and the Radar Index is calculated on the whole set.
Spot
The prompts where competitors get ahead rise to the top, together with the domains feeding their citations.
Publish
Targeted content is produced and goes live on the network, server-side and without external sign-off.
Measure again
The new reading is compared with the baseline, AI traffic is read page by page, and the plan is adjusted.
A cadence that holds
Baseline measurement. The first reading is frozen and becomes the comparison point for every measurement that follows.
Automatic full reading. All your prompts, all the engines, at the same point in the week, so the curves stay comparable.
One-off reading. Just before and just after a publication, a product launch or a move from a competitor.
AI traffic. The pixel records every crawler visit to your published pages, without waiting for the next reading.
The team
The articles on the network are written by writers
The platform records, sorts and compares, but it does not write on your behalf: behind every piece published on the network, somebody chose the angle, checked the claim and signed the text themselves.
The person who opens your account is the one who will still have it in two years. Our teams sit in Mougins and Aix-en-Provence, well away from Paris turnover.
An in-house newsroom
Our writers work the network's subjects all year round, so a piece on barbecues or on sensitive skin comes out with the vocabulary of the field.
A contact who stays
The person who opens your account is the one who will still have it in two years. Our teams sit in Mougins and Aix-en-Provence, well away from Paris turnover.
Twenty years in the trade
We have been publishing sites and buying media since 2006, which gives us useful perspective when a new channel starts to reshape how products get discovered.
Our commitments
You can check what we claim
Assistant answers are not advertising slots, so progress is built through content rather than bought. Our commitment covers the mechanism, and it can be checked week after week.
A reproducible reading
The same protocol every week, multiple repetitions, raw answers stored and readable prompt by prompt.
An open calculation
The six Radar Index axes and their weightings are published, so you can redo the calculation yourself.
Engines kept apart
Assistants that browse the live web and those answering from training are tracked separately, which shows where the progress is actually happening.
Dated production
Every published article carries its date, its target prompt and its host site, with AI traffic recorded page by page.
Frequently asked questions
The questions that come up most
Since 2006
Twenty years of building audiences
Georadar extends a publishing and acquisition business we have run for twenty years, with the same sites and the same teams, so the platform rests on a proven inventory rather than on a network assembled for the occasion.
Published by
Georadar is published by R-Advertising
R-Advertising publishes the 70-site thematic network exposed in the platform and has been buying media since 2006, so measurement and activation come from the same operator. There is no media house to coordinate and no middleman to convince between the decision and going live.
Your first measurement can run today
Creating the account takes a minute and asks for no card. You approve your prompt plan, launch the reading, and the score appears as soon as it finishes.