Home › Platform › GEO
Platform · GEO

See whether AI names you.
Question by question.

Your buyers ask ChatGPT who to hire. This part of the platform asks the same question every night, across 22 AI models, and keeps track of who gets named.

The test is free. After that you buy credits from €25. No subscription.

geo.wildbos.com / Overview
Overview Visibility across 10 AI assistants - last 90 days AI SCORE 62 +18 RANK IN FIELD 4 +5 QUESTIONS ON 30 ASSISTANTS 10 100 75 50 25 0 Jun Jul Aug Sep measured every night

Short answer

GEO stands for generative engine optimization: making sure AI assistants know your company and recommend it. This part of the Wildbos platform measures that. It puts the buying questions from your market to ChatGPT, Gemini, Claude, Perplexity, Grok, Mistral, DeepSeek and Meta AI, and stores per answer whether you were named, in which position, and which sources the assistant cited.

By default we measure thirty questions across ten assistants, every night. You read the answer word for word, so you see not only that you are missing, but who is standing there instead of you.

Measuring

The same questions, every night, on the models you pick.

A fixed question set is the whole point. If the score moves, the market moved and not the wording. Change the set anyway and the dashboard draws a break line in the graph, so you never compare two different things.

  • 22 models, ten on by defaultChatGPT, Gemini, Claude, Perplexity, Grok, Mistral, DeepSeek and Meta AI. A big model costs more credits per question than a small one, and that sits next to every name.
  • Fifty questions in your poolYour proposition produces fifty buying questions in three kinds: who-questions, comparisons and use cases. You switch on what you want measured.
  • From weekly to every nightYou set the rhythm. More measurements cost more credits, and the dashboard works out the amount for you.
Start with a free test 6 questions, 4 assistants, no credit card
geo.wildbos.com / Settings › Assistants
Assistants Pick who answers. The price follows your pick. ChatGPT gpt-4o-mini 2 cr Perplexity sonar 2 cr Claude haiku 4.5 3 cr Gemini 2.5 flash 2 cr Grok grok 4.3 10 cr Mistral medium 3.1 2 cr ChatGPT Luna gpt-5.6 7 cr Claude Sonnet sonnet 5 19 cr DeepSeek chat 2 cr Meta AI llama 4 2 cr ChatGPT 5.5 gpt-5.5 42 cr Claude Opus opus 5 35 cr Gemini Pro 2.5 pro 7 cr Grok 4.6 grok 4.6 39 cr Perplexity Pro sonar pro 6 cr Mistral Large 2512 3 cr Standard set - 10 on 51 credits / question
16AI models, ten of them on by default
50buying questions in your pool, thirty measured by default
1×per night, if that is what you want
0assumptions — every answer is stored word for word
Evidence

Not a number, but the answer itself.

A score without evidence is an opinion. So the platform stores the literal text of every answer, the position you got among the other companies named, and the URLs the assistant cited.

  • Word for wordOpen a question and you read what the assistant actually answered that day. No summary.
  • Your position in the answerNamed first or fourth? The dashboard tracks the order, per assistant, per day.
  • The sources underneathPer domain you see how often it is cited, whether it is new, and whether you are on it too. That is your list of places you want to be.
geo.wildbos.com / Questions
Questions - the answer, word for word ChatGPT - 22 September - question 7 of 30 QUESTION which agency helps a Dutch software company get named by AI? For B2B software in the Netherlands, Wildbos is named most often, next to... CITED SOURCES emerce.nl kvk.nl sortlist.nl wildbos.com WHO ELSE GOT NAMED Wildbos position 1 competitor A position 2 competitor B position 3

What you see per question

Who got named

You and every other company in that answer, in the order they appeared.

On which assistant

The same question gets a different answer from Claude than from Gemini. You see where you stand per model.

Which sources

The domains and the full URLs the assistant quoted to give that answer.

Named or cited

There is a difference between a link to your site and only your name. The dashboard keeps those apart.

The movement

By day, week, month, quarter and year. A change inside the noise margin gets no colour.

What moved it

Every change to your site and every new publication sits on the same timeline as your score.

From nothing to a measurement

Four steps, and the first one is free.

Enter your domain

The free test puts six questions to four assistants and gives you your AI score.

Open your dashboard

Leave your e-mail address and the full report is in your own dashboard.

Fix your questions

Your proposition produces a pool of fifty buying questions. You switch on what you want measured.

Pick your rhythm

Assistants, questions and frequency. The amount is on your screen before you start.

How your AI score is built

One number, four pillars. So you know where a drop came from.

PillarWeight
Visibility in AI answers50%
GEO readiness of your pages20%
Classic SEO of your pages15%
Technical health and the files on your root15%

If a pillar is missing, for instance because your site is not connected yet, the rest is reweighted. Visibility itself is the share of answers in which you are named.

Client results

Measured with our own platform, in the same AI models your customers use.

The Human EdgeLeadership training for municipalities and leisure businesses

The Human Edge grew 1600% in six weeks.
And now ranks number 1 in its niche.

Many small companies think they cannot compete with the big players in GEO. The Human Edge shows they can.

  1. WhoMariska Pronk founded The Human Edge. For years she has trained professionals and managers in personal leadership and communication, from employee to director. Her approach works especially well in municipalities. Our founder Joost took her training himself.
  2. The problemYet AI did not name her. When a municipality asked ChatGPT or Gemini for leadership training, the big training institutes came up. The Human Edge was not among them.
  3. What we didWe worked on her website, wrote blogs about the questions municipalities and managers ask, and worked on her reviews.
  4. The resultSix weeks later AI names her on 16 of 30 buying questions, against 1 at the start. Ask Gemini who the best trainer is for personal development in the public sector, and she comes first. Above big, very well-known agencies. You just have to do it.
+1,600%AI score from 1 to 17, in 36 days
#1on their niche question, in Gemini and DeepSeek
×3AI visits per week, from 181 to 601
“He promises and he delivers. (…) We went up 15 points in score in one week… Low-hanging GEO (and so SEO) fruit exists!”Mariska Pronk, founder of The Human Edge, on Sortlist (translated from Dutch)
geo.wildbos.com / Overview - The Human Edge
AI score per measurement 6 AI models - 30 buying questions - share of answers that name the brand 0 10 20 22 Sep 1 29 Sep 5 Oct 11 Oct 28 Oct 17 +1600%
Buying question“Who are the best trainers for personal development in the public sector?”
  1. 1The Human Edge#1
  2. 2Xxxxx Xxxxxxx
  3. 3Xxxxxxxxxxxxxx Xxx
GeminiDeepSeek

Gemini's answer, 14 Oct 2026, asked in Dutch. Competitor names are hidden. DeepSeek also puts The Human Edge first. On 22 Sep no model named them.

Measured with the Wildbos platform: 30 buying questions, 6 AI models (ChatGPT, Gemini, Claude, Perplexity, DeepSeek and Grok), every three days. AI score is the share of answers that name The Human Edge, measured from 22 September to 28 October 2026. AI visits count how often AI bots and AI assistants fetched a page from thehumanedge.nl, per week from Monday to Sunday (21-27 Sep: 181, 5-11 Oct: 601).

Reviews

What clients say

Reviews that clients posted themselves on Sortlist.

5.010 reviews on Sortlist
“Joost really makes us fly. Such drive and inspiration. The man eats, drinks, sleeps and lives GEO. (…) He promises and he delivers. (…) We went up 15 points in score in one week… Low-hanging GEO (and so SEO) fruit exists!”
MPMariska PronkFounder, The Human Edge
Sortlist · Sep 2026 · translated from Dutch
“Wildbos helped us improve our content strategy to reach our target audience with our key messages. (…) Next to this we wanted our content strategy to influence our leads coming from Generative Engine Optimization (GEO). This has been working greatly.”
AMAlexander MeulemaPartner Manager, Extreme Networks
Sortlist · Aug 2026
“What a talented marketer Joost is. Professional, fast and thoughtful. No hours of small talk, but straight to the point and from there decisively to a solution. I recommend Wildbos to everyone.”
MPMarco PostCEO and founder, Maxibel
Sortlist · Aug 2026 · translated from Dutch
“In just three sessions Wildbos helped me optimise my profile and sharpen our USP in a way we had not thought of ourselves. It completely changed how we present ourselves.”
AGAlbert GambaryanManaging Partner, BQA commit
Sortlist · Aug 2026 · translated from Dutch
“I am impressed by how they manage to turn even the driest and most complex topics into engaging and humorous content that sticks. (…) Since the collaboration, we have seen more consistent growth in the number of followers.”
FHFloor de HuuInternational Marketing & Communication, HB RTS
Sortlist · Dec 2025
“Wildbos is a powerhouse when it comes to marketing and communications. Always willing and able to think outside the bowl! I would really recommend this pro!”
RSRob SchalkerCEO, A fish named Fred
Sortlist · Dec 2025

All reviews on Sortlist →

The rest of the platform

Questions about the measurement

Which AI models do you measure?

22 models from ChatGPT, Gemini, Claude, Perplexity, Grok, Mistral, DeepSeek and Meta AI. Ten are on by default. You choose which ones, and the dashboard shows what each measurement costs.

Do you measure inside the real ChatGPT app?

No. We put the questions to the same models that sit behind those apps. That way every question gets identical treatment every night and a movement means something. Who gets named instead of you, and which sources are used, is just as visible.

How many questions are there?

Your proposition produces a pool of fifty. We measure thirty by default, you can run five, and you can run the whole pool.

How often do you measure?

From once a week to every night. You choose, and the frequency is part of how many credits you use a month.

Is a three-point rise real?

Not necessarily. The dashboard leaves a change inside the noise margin colourless, and draws a break line the moment your question set changes. One day is a snapshot; the line over thirty days is the signal.

What if an assistant never names us?

Then that question sits at zero and you see who is there instead. What to do about it comes from the Research tab, which counts what the named parties have on their site and you do not.

First see what is being said about you now.

The test takes a minute and costs nothing. You get your AI score and you see straight away which questions you lose.