A candidate researching jobs no longer opens ten tabs. They ask ChatGPT one question and read one answer.
New figures show just how fast that shift is happening now. SE Ranking tracked 101,574 websites between January 2025 and April 2026. Traffic from AI search engines grew 16-fold during that period, from 0.02 to 0.32 per cent of all web traffic. ChatGPT accounts for almost 75 per cent of that; Gemini and Perplexity share what is left.
That seems small, until you see what lies beneath it. BrightEdge measures that Google now shows its AI Overviews above almost half of all search queries, 58 percent more than a year earlier. And on Google AI Mode, 93 percent of search queries end without a single click.
What does this mean for you as an employer?
It means two things at once. The first: your careers site gets fewer visitors, and that's not going back up. The click that used to come naturally is now intercepted by the AI answer. What the model says is what the candidate reads.
The second is less well known, and sharper. The candidate who does click through from an AI answer is more deeply engaged. The same metric: visitors from AI stay on a site for an average of 9 minutes and 19 seconds, compared with 5 minutes and 33 seconds from ordinary search results. That is 68 percent more attention.
The side that is shrinking is the number of clicks. The side that is growing is the value of who is still coming. Whoever arrives via AI has already read the answer, has already compared you, and consciously chooses you.
That is the core of it. The orientation takes place within the model. The selection too. By the time someone lands on your site, the first impression has already been formed by a system you did not write.
And that system does not name ten employers. It names two or three.
And that doesn't just affect candidates. A client looking for a supplier, a journalist checking you out, a student choosing an internship: they're all asking the same question to the same model. That same invisibility costs you every single one of them.
Try it out yourself
Ask the question yourself. “Which healthcare employers have a good culture?” “Where can a data engineer in the Netherlands really progress?” The answer comes in seconds, with names included. Are you among those names, or not.
This is where most employers drop the ball. They measure their employer brand by reach, by views, by applications. Nobody measures what a language model says about them to someone who doesn't know them yet. And right there, at the exploratory question of a stranger, the choice is made.
Being mentioned depends on one thing: content that a model can read, understand and trust. Clear pages, factual answers, a recognisable source behind them. No advertising budget buys you that spot. What is vague, the model skips. What is missing, does not exist.
That last point is the hard rule of this profession. What a model cannot read, it cannot cite. No matter how good your story is: if it is not in a format that AI picks up, it falls outside the answer. And outside the answer means out of the picture.
For a language model, there are now only two types of employers. The employer with clear, consistent, readable information, which is structurally recommended. And the employer with thin or outdated pages, who isn't even mentioned. There is hardly any middle ground between the two; you are in the answer, or you are not.
What is still an advantage today will be the norm in six months. What still feels optional today will be the basics in a year. And what you aren't measuring today will cost you candidates in eighteen months that you never saw coming.
We see it reflected in the work. For a major job board site that we monitored over 47 days across five AI platforms, comprising over 9,300 answers, the average AI visibility came out at 71.8 per cent, with its own domain as the most cited source and less than 1 per cent negative sentiment. At the baseline measurement prior to that, it stood at a fraction of that. The difference came from technology and structure that AI could finally read. Not a single word of the story itself had been changed.
That is the good news in these figures. The shift is measurable, and therefore steerable. You can see what AI is saying about you today, which questions you are missing on, and which competitor is taking your place. And you can adjust that before it costs you a candidate.
The first step is not a big one. Measure what AI currently says about you, based on the questions your target audience is actually asking, across the platforms where they ask them. That is a baseline measurement spanning a few weeks. And it yields something most employers have never seen before: an honest picture of their own place in the answer.
This is already happening. Every day, in conversations you don't see. The only choice you have left: knowing what is being said about you there, or leaving it to chance.
My name is Bryan Peereboom, founder of AI Rebels. We make employers visible in language models.