For two months, we tracked how a brand appears in AI responses. The figures show why optimising for ‘the AI’ doesn't exist, and where your employer brand has its biggest blind spot.
✦ IN BRIEF
| Does your brand appear in AI responses? That depends entirely on which AI you’re using. In our monitoring of a single brand over two months, visibility ranged from 19% on ChatGPT to 69% on Google AI Mode. The same brand, the same week. Anyone optimising for ‘the AI’ without knowing which one is optimising for nothing. |
✦ THE MOMENT THAT COUNTS
“Which employer in my sector is the best to work for?”
A candidate opens ChatGPT. No Google, no LinkedIn. An answer within ten seconds, with names, positions and recommendations. For a growing portion of talent, that's their first impression of you as an employer.
Does that answer include you?
Most organisations don't know this. Not because they're doing a bad job, but because they never measure it. We do. Per platform, per question.
✦ THE WHITE SPOT
Almost no one measures if you show up
There is a lot of research into how AI search engines scan the web, but hardly anyone measures whether your brand appears in the answer, let alone answers the question a candidate asks. That AI splits one question into multiple searches, known as query fan-out, is well-documented. Who comes out on top per platform remains a blind spot.
So we did something that rarely happens. We followed one brand for two months, logging every AI response to see if it appeared, in what position, and with what sentiment.
THE MEASUREMENT
One brand, two months, 610 AI answers
| We monitored one brand from 7 April to 2 June 2026 across five AI platforms: ChatGPT, Google AI, Microsoft Copilot, Gemini, and Claude. 8 prompts, 610 AI responses, measured weekly. For each response, we recorded: did the brand appear, in what position, and how was it described? The brand and sector have been anonymised. The patterns are what count. |
The key finding in a nutshell: the same brand was three times more visible on one platform than on the other. 69% on Google AI Mode, 19% on ChatGPT. No difference in the brand. Just a difference in which AI you happen to open.
The Google ecosystem worked in favour of this brand. ChatGPT and Claude, precisely the platforms recruiters and HR managers use daily, lagged behind. Optimising for ‘the AI’ therefore means nothing. You are optimising for six different search engines simultaneously, each with its own outcome.
✦ FINDING 1
The crawlers found everything. They didn't cite it.
AI crawlers visited the site almost daily. Yet, the most intensely crawled pages were not among the most cited. Access is not the bottleneck. Citeability is.
The crawlers find your content effortlessly. Whether a language model then picks up that content depends on how citable each block is, not on whether the bot gets in. Pages that are not structured as clear, standalone answers will be ignored, no matter how often they are crawled.
✦ FINDING 2
Each platform draws from a different, sometimes minuscule, number of sources.
The number of sources a platform cites varies enormously. In our measurement, Google AI Mode quoted 1,014 times from 146 domains, while Claude quoted once from one domain in the same period.
| Platform | Citations | Unique domains |
|---|---|---|
| Google AI Mode | 1.014 | 146 |
| Microsoft Copilot | 711 | 104 |
| Google AI Overview | 461 | 102 |
| ChatGPT | 205 | 78 |
| Claude | 1 | 1 |
The smaller the platform quotes, the fiercer the battle for that one spot. On Claude, one source was enough to determine the whole picture. If you're not there, you don't exist.
✦ FINDING 3
More often mentioned is not described warmer
Of the 340 mentions, 0% was negative: 79% was neutral, and 21% was positive. However, whilst visibility increased, the sentiment score fell slightly. The brand was seen more often and, at the same time, described in more general terms.
Visibility and appreciation are not the same. Being mentioned is step one. Being recommended is step two, and that doesn't come from more of your own content, but from external validation: customer cases, reviews and mentions from third-party sources.
The blind spot
The candidate's question wasn't even addressed.
Herein lies the biggest lesson for your employer brand. Of the eight questions measured, not one was about the candidate. They all revolved around the business decision. The question a candidate asks wasn't even on the radar.
In comparison questions, this brand was by far the most visible, ranking #1. On the candidate side: nothing measured, so nothing managed. That is precisely where employer branding comes into its own, and precisely where most brands are blind. Not because their story is poor, but because nobody measures it and nobody creates the content for it.
✦ THE CONCLUSION
You can't optimise for ‘the AI’
There isn't one AI. There are six search engines with six outcomes, each with its own set of sources and its own flavour. On one platform you're the reference, on another invisible. And on the question that matters most for talent, almost nobody measures.
Anyone who takes their employer brand seriously in the AI era, therefore, does not start with more content. They start by measuring, per platform, and then build strategically.
✦ OUR APPROACH
How AI Rebels are gaining a grip
Measuring. With our services, we look at which questions your sector raises per platform and whether you appear in the answer. ChatGPT, Gemini, Claude, Copilot and Google AI, measured individually, week after week.
Building. We make your content quotable: standalone answer blocks that literally answer the question, E-E-A-T pages, schema markup, and open crawler access. And we cover the candidate side, not just the business side.
Strengthen. We focus on the external sources from which AI draws: reviews, rankings, and mentions. That's what turns neutral mentions into active recommendations.
Three questions to ask today
- Ask your sector-specific question to ChatGPT and Claude. Do you get the same answer from both, or do you get a response from one but not the other?
- Question: “Is [jouw bedrijf] a good employer?” What is the source of the answer?
- Who measures the candidate question with you, or does no one measure it?
✦ GETTING STARTED
What you are doing this week, month, and this quarter
This week
- Ask your main sector questions to all five platforms and record for each platform whether you appear.
- Bouw een vergelijkingspagina, “jij versus de markt”, als direct vraag-antwoord.
This month
- Restructure your most crawled pages into a citable Q&A format.
- Publish candidate-focused content: culture, growth opportunities and employee stories, with the real candidate question as the headline.
This quarter
- Generate external validation: reviews and mentions on third-party sources.
- Implement a structured survey per platform, with a broader set of 30 to 50 questions.
What now?
Measure first. Then build.
The shift is already underway. Candidates are asking AI, and AI answers with or without you, differently on each platform. We measure where you stand and ensure you appear.
Would you like to see what AI says about you today, per platform?Plan a conversation here |