For two decades, B2B discovery ran through intermediaries. A buyer searched *best [category] software*, landed on a review grid or a *top 10* roundup, and picked three names from it. Those intermediaries were powerful because they owned the click — they sat between the query and the vendor, and everyone paid rent to be there, in listing fees, review campaigns, or sponsored placement.
That mechanism is changing, and it's worth being precise about how. The intermediaries aren't disappearing. Their role is shifting from destination to source — and that changes who you have to influence, and how.
What's actually happening to the click
The measurable part is what happens on a results page with an AI summary on it. When Google shows one, only 8% of users click any traditional link — versus 15% without one — and just 1% click a source inside the summary itself (Pew Research, 2025).
Sit with those numbers from a listicle's point of view. Its entire business model assumed the click. If a buyer's question gets answered above the results without one, the roundup can still be *read* — by the model — but it stops being *visited*. Its content still shapes the answer. It just no longer collects the traffic that made it valuable to be listed in.
Meanwhile traffic is genuinely arriving from the new direction: retail visits from generative-AI sources grew 1,200% between mid-2024 and early 2025 (Adobe Analytics), and those visitors convert better than traditional organic (Semrush, 2025).
Source, not destination
Here's the shift in one line: you used to want to be listed somewhere a buyer would visit. Now you want to be described somewhere a model will read.
This sounds like a small distinction and it isn't, because it changes what "being on the list" is worth. A prominent placement on a page nobody visits still has value — the model reads that page — but the value is now *indirect and unattributable*. You can't see it in referral traffic, and you can't A/B test it. It shows up as being named in an answer somewhere you'll never observe.
It also changes the *shape* of what wins. A review grid position is a rank. An answer is a synthesis assembled from several sources — which means a company mentioned consistently and specifically across many places can outperform one ranked highly in a single directory. Corroboration beats placement.
What replaces the intermediary
Three things, roughly in order of how much control you have:
Your own pages, if they're extractable. The model needs something quotable and verifiable about you. Pages that answer real buying questions with named prices and specific numbers are the closest thing to a direct line you have — the technical requirements are here.
Corroboration across the open web. The mentions, comparisons, discussions, and reviews the model cross-checks. This is the descendant of the old link-building and PR motion, with a different success criterion: not a backlink for authority, but a *statement about you* a model can read and confirm.
The intermediaries themselves, still — as sources. Directories and roundups remain worth being in, because they're exactly the kind of corroborating source models retrieve from. What's changed is why: not for their traffic, but for their testimony. That reframes how much they're worth paying for, and it's a calculation most marketing budgets haven't redone.
What this means for your budget
Two adjustments, neither of them dramatic:
Stop valuing directory placements purely on referral traffic. If you're measuring a listing by clicks it sends, you'll under-value one models read and over-value one that merely converts well. Judge placements partly on whether they're the kind of source an assistant would cite.
Move some spend from renting placement to owning statements. A page of yours that answers a buying question with quotable specifics is an asset that compounds; a listing is rent. The old ratio between those two made sense when the intermediary owned the click. It makes less sense now.
The honest uncertainty
Nobody can tell you exactly how fast this is happening in your category, and anyone who claims a precise number is guessing. What's measurable is that the click-through economics behind the old model are eroding (Pew, 2025) and that referrals from the new one are growing fast (Adobe, 2025). Whether your buyers are 5% or 40% through the transition is an empirical question about *your* category, and it's answerable.
That's the actual first move: measure it rather than theorise about it. Ask the assistants your category's buying questions and see whether they have a confident answer, and whose name is in it — five minutes gets you a first read, and the free citation audit runs the full panel and scores it.
If the answer is that nobody in your category is being recommended yet, the intermediary hasn't been replaced — the position is simply vacant. That's the most valuable finding available, and it has a shelf life. Ascent is the program we run against exactly that window.
