What GenAI Means for Generic Keyword Advertising — And Why the Funnel Is Quietly Relocating

For most of the history of paid search, generic keywords did a job that was easy to underrate. They were a proxy for intent we could not otherwise observe. When someone searched "best running shoes for flat feet" or "accounting software for a small team," they were not making a decision yet — they were researching. Advertisers paid to be present in that research moment, on the reasonable bet that they could inform an undecided buyer before a competitor did.

The model rested on a simple assumption: that the research happens inside a search box. I think that assumption is now worth re-examining, and the implications go a little deeper than the usual "AI versus SEO" conversation suggests.

When someone opens an AI assistant and asks which accounting tool suits a five-person agency, a meaningful part of the journey has already moved. The shortlisting, the comparison, the weighing of trade-offs increasingly happens inside a system that runs no ad auction. The broad, exploratory query that fed the top of the paid-search funnel for two decades is, in some categories, less likely to land on a page where a bid can compete for it.

What tends to remain in the search box is the more decisive end of the journey — the branded query, the price check, the final comparison, the clear purchase intent. These searches do not disappear; if anything they become more valuable, because the person typing them has already done much of their thinking. But there may be fewer of them, and they are likely to be more contested.

This is where I would be careful reading the dashboard. For years, success in generic-keyword campaigns was measured partly through cheap, broad volume — large numbers of impressions and clicks at a tolerable cost. As assistants absorb more of the research phase, some of that volume may erode. The natural reflex is to treat this as a performance problem to optimise away. My own view is that part of it is structural: the cheap top-of-funnel layer is gradually moving into an environment where, at least for now, there is no position to buy.

That is the part I find genuinely interesting rather than alarming. The top of the funnel has not vanished — in some categories it is relocating to a place advertisers cannot yet bid into. If the moment of consideration increasingly happens inside a model that synthesises an answer rather than listing ten links, the operative question shifts. It is less "which keyword do I buy?" and more "is my brand one the model surfaces when nobody typed my name?"

That is a harder and slower thing to influence, and I think it rewards the fundamentals. Being widely and consistently referenced, credibly reviewed, and clearly described across the open web is the kind of presence these systems tend to draw on. It favours brands that built genuine reputation over those that simply rented visibility. The emphasis moves from purchasing attention toward earning recommendation — related skills, but not identical ones.

I want to be measured here. None of this means search advertising is in decline. Transactional intent still flows through search and will continue to for a long time; people compare prices, find local suppliers, and complete purchases they have already decided on. What I am describing is narrower: a gradual softening of the broad, generic, research-stage keyword as a dependable engine of cheap reach. It will play out unevenly, category by category, and will be easy to dismiss in any single quarter.

So I would simply be cautious about assuming generic-keyword volume returns indefinitely to past levels in every category. The behaviour that produced that volume appears to be shifting, and customer behaviour rarely shifts back on its own.

The question I keep returning to is this: if the research moment increasingly happens somewhere you cannot bid, what are you doing today to make sure your brand is the one that surfaces when nobody typed your name?

The above reflects my personal views only and is intended for informational and discussion purposes. It does not represent the position of any employer or organisation.