A leadership team somewhere is about to look at a chart and reach the wrong conclusion. The chart is accurate. The conclusion isn’t.
Start with what changed.
Forrester asked business buyers where they go to research a purchase. Twice as many named generative AI or conversational search as their most meaningful source of information than named any other source. Ahead of vendor websites. Ahead of product experts. Ahead of salespeople. And the share of buyers using AI somewhere in the process reached ninety-four percent.
That isn’t a forecast. It’s a description of last year.
Two findings sit underneath the headline and both matter more than it does.
The first is that most buyers arrive with a favourite already chosen. Sixty-eight percent have a front-runner in mind before the purchase process formally begins.
Before it begins. Not partway through, once the field has been walked and the demos sat through.
The second finding is what happens to that front-runner. It wins about eighty percent of the time.
Sit with what that does to the rest of the process. The demo. The proposal. The reference call. The pricing negotiation everyone spends a fortnight preparing for. A good deal of it is confirmation of a view that formed weeks earlier, at somebody’s desk, in a chat window, in about four minutes.
Sellers have suspected as much for years. What’s new is where the view forms.
Now the number that will actually get discussed.
B2B organic traffic is down somewhere between ten and forty percent over the past year. It’s on every dashboard. Visible, countable, falling.
It’s going to be read as underperformance. In some companies, budgets will move because of it.
That reading is wrong, and the reason is worth slowing down for.
For about twenty years, website traffic worked as a stand-in for commercial interest. Somebody had a problem, searched, arrived, read. The visit was never the interest itself. But it tracked closely enough that treating one as evidence of the other almost never caused trouble. Measurement systems got built on that. So did org charts. So did careers.
The relationship has broken. The number survived.
Which is a pattern, and it has very little to do with AI.
An organisation adopts a proxy because the thing it actually cares about is hard to observe. Time passes. People stop remembering it was ever a proxy. Then conditions shift, the link quietly dissolves, and everyone goes on reading the number with the same confidence they had when it worked.
Broken measurements don’t announce themselves. They keep producing figures. That’s what makes them expensive. A metric that has stopped working looks identical to a metric delivering bad news.
Here the gap is unusually wide. Interest hasn’t fallen at all. Buyers are doing more research than before, earlier than before, and more of it on their own. Every bit of that consideration is still happening. It just stopped passing through a page anyone owns, instruments, or reports on.
One more feature deserves attention, because it’s genuinely new.
There is no sponsored slot inside an AI recommendation.
You can’t buy the shortlist. Presence in those answers is a function of what exists about a company. How much of it, in how many places, in enough detail that a model has something to cite. That isn’t media buying. It’s closer to reputation. And reputation has never responded well to being bought in a hurry.
Which makes this a harder problem than the one it replaced. Not an easier one.
So what should the chart mean?
It’s a measure of one route to consideration, read at a moment when most of the traffic took another. Still worth looking at. Just no longer worth looking at the old way.
The companies that get dropped from a shortlist this year won’t be told. They’ll find out from a pipeline number at the end of the quarter. And most of them will blame something else.
Photo by Sergey Meshkov: https://www.pexels.com/photo/miniature-shopping-cart-8481889/
Leave a Reply