68 per cent of all Google searches ended this spring without a single click. SparkToro measured this from January to April, using clickstream data from Similarweb. Two years earlier, the figure stood at 60 per cent. The direction is clear.
Zoom in even further on what that figure does to you as an employer.
A candidate asks a question. They get an answer at the top of the screen. And they leave without clicking anywhere. Your careers site, which you've spent a year working on, doesn't even enter the picture in that moment.
Searching has become a conversation
In the past, searching was a passage. You typed something, you got ten blue links, you chose one. The click was the whole game. Today, searching is a conversation. Google’s AI Mode passed the threshold of 1 billion users a month in May, and the volume doubles every quarter. The answer is right there immediately. The links underneath have become a sideline.
Pew Research laid it bare. If there is an AI summary above the results, 8 per cent of people still click through to an ordinary link. Without that summary, it is 15 per cent. Almost a halving, in one fell swoop.
For recruitment, this hits at the core of how candidates find you.
Think about the questions a jobseeker is actually asking. “Which employer in logistics treats its people decently?” “Where can I progress as a data engineer without working myself to the bone?” “Which company suits someone with my background?” Those are whole conversations. And the model answers them with company names.
The outcome is binary. Your name is in that answer, or it is not.
ProRail case
At ProRail, we saw precisely this happen. Candidates asked language models about working in the rail sector, and the answers often focused first on the well-known names around it. We redesigned the FAQ architecture on werkenbijprorail.nl, directly based on the questions candidates ask AI, as measured via PromptWatch. Consolidation of URLs, schema, an llms.txt telling models what is there. Boring interventions. Exactly why they work.
A language model does not quote an atmosphere. It quotes what is clear, what is consistent, and what it can trust. A carousel of smiling colleagues means nothing to a model. A clear answer to a real question means everything.
what being quotable concretely means
Being quotable is more concrete work than it sounds. It means you literally answer candidates' questions on your own page, in plain language, without a marketing layer over it. It means that information about salary, career progression, and culture is in one place and not spread across five sub-sites that contradict each other. And it means you supply that information in such a way that a model can read it, match it to a question, and pass it on with peace of mind. Technology and honesty, in the same motion.
Here lies the flaw in reasoning I see in far too many employer branding teams. They invest in the most beautiful final destination, while the journey never takes place anymore. The candidate no longer arrives at your page. They get a summary of who you are, written by a model that has never met you, built from whatever it happened to find out about you.
What is missing will be filled in. What is unclear will be simplified. And what is nowhere stated does not exist.
The latter is the sharpest. If your story is not readable for a model, the model writes your story for you. Or it leaves you out. Both are worse than a weak job advert, because you don't see it happening. There is no rejection, no click count that drops. You simply aren't there.
Why you don't see it happening
Most organisations only notice this when it has already been going on for months. Applications slowly drop off. The agency gets the blame, or the labour market, or the season. Meanwhile, something more fundamental has shifted: the conversation in which candidates make their choice is taking place in a space where you do not have a measurable presence.
I understand this feels uncomfortable. You've just renewed your careers site, sharpened your EVP, got your socials in order. That work still counts. It just no longer carries your findability on its own. The findability you used to buy through Google with SEO, you now earn anew within the answers of AI. That's a different discipline, with different rules.
What can you do now?
There is also good news: it is early days. Most employers are not doing anything about this yet. Anyone who ensures their story is citable now will be featured in the answers later while the competition is still waiting for better click-through rates that are no longer coming.
What is still an advantage today will be the norm in a year's time. What still feels optional today will be the baseline in eighteen months. And those who start today will help determine what AI has to say across the entire sector.
Start small. See what questions candidates are actually asking AI about your kind of work. Test what ChatGPT, Gemini and Perplexity are saying about you today. There is a strong chance the answer will surprise you, and not in a pleasant way.
That conversation is going ahead, with or without you in it.
My name is Bryan Peereboom, founder of AI Rebels. We make employers visible in language models.