A founder I talked to recently said his demos had started to feel strange. The prospects showed up already knowing things. They had opinions about his pricing before he quoted it. They compared him to two competitors he had not mentioned. A few had already decided he was the “cheaper, lighter” option before the call began. He could not work out where any of it came from, because none of it came from him.
It came from a chat he was never part of. And it is happening to every B2B SaaS company right now, whether or not they can see it.
The top of the funnel moved, and it moved somewhere you cannot watch
For twenty years the buyer journey started with a search. Someone typed a problem into Google, clicked through a few results, landed on your site, and entered your funnel where your analytics could see them. You optimized that path. You measured it. You knew roughly where people came from and what they read.
That first step is now happening inside an AI conversation. Before a buyer ever visits your site, they ask ChatGPT or Perplexity what the best option in your category is. They ask which tool fits a team their size. They ask what your product is weak at, what it costs, what people switch to it from. They get a clean, confident answer in seconds, and they form a working opinion of you before you know they exist.
The research that used to happen across ten browser tabs, where you could at least show up, now happens in one chat window where you cannot. The top of your funnel did not shrink. It relocated to a room you are not allowed into.
What actually happens in that chat
It is worth being specific about what the engine does to you in those few seconds, because it is more than a mention.
First it builds a shortlist. The buyer asks for the best option and the engine returns three or four names. If you are not one of them, you do not exist for that buyer, and you will never see the impression you lost.
Then it assigns each name a role. Not just “here are the options,” but “this one is the industry standard, this one is built for scale, this one is fine if budget is tight.” The engine decides which sentence sits next to your name, and that sentence frames everything that follows. Being on the list is not the same as being the recommendation, which is a gap most teams cannot even see on their own dashboard. I wrote about that gap in getting cited is not the win you think it is.
Then it anchors a price perception, a “you switch to it from” story, a sense of who you are for. All of it assembled from whatever sources the engine trusts on that question, which for most categories means review sites, forums, and your competitors’ comparison pages. Your own site is often not even in the mix. The buyer arrives carrying a version of your product that other people wrote, and the fix for that lives in how to change the way AI search describes your brand.
Why your analytics will never show you this
Here is the part that makes it genuinely hard to manage. The most important moment in the new buyer journey leaves almost no trace.
When a buyer finally does click through, often they arrive with no referrer, because the click came out of an app or a pasted link. Your analytics file the visit under Direct, right next to people who typed your URL from memory. The chat that shaped the entire decision shows up as nothing, or as a Direct visit you cannot explain. The deeper mechanics of that tracking gap are in how to track AI search traffic in GA4.
So you end up in a strange position. The single biggest influence on your pipeline is a conversation you cannot see, cannot measure cleanly, and did not get to take part in. Most teams respond by optimizing the parts they can still see, the landing page, the demo, the email follow-up, and quietly wonder why none of it moves the number.
The demo is not discovery anymore
The practical effect is that the sales call has changed jobs. It used to be where the buyer learned who you were. Now a lot of it is correction. You are not making a first impression, you are managing one that a machine already made on your behalf, often weeks before the call.
That is a worse position to start from than no impression at all, because a wrong frame is stickier than a blank slate. If the engine told a buyer you were the lightweight option, your demo now has to overturn a belief instead of forming one. You are spending the call climbing out of a hole you did not dig and could not see.
What you can actually do about it
You cannot be in every chat. You cannot watch them or measure them directly. What you can do is change what the engine has to work with when your category comes up.
The engine assembles its answer from sources. If the clearest, most specific, most verifiable source on your category’s buying questions is yours, you stop being at the mercy of a competitor’s comparison page. That is the entire job: figure out the exact questions your buyers ask an AI, find where the current answers leave you absent or mislabeled, and become the source the engine would rather quote. The starting point is mapping those questions, which is the buyer query map, and the way to know where you actually stand today is to measure your AI search visibility before you assume anything.
None of that puts you in the chat. It changes what the chat says about you when it happens, which is the only lever you actually have.
Why this matters
The companies that win the next few years of B2B software will not necessarily have the best demo or the sharpest landing page. Those still matter, but they happen too late. The buyer has already been to the conversation that mattered, and arrived at your door with a verdict.
The real contest moved upstream, into a chat you will never see, decided by which sources the engine trusts on the questions your buyers actually ask. You do not get to attend. You only get to influence what is on the record before the buyer walks in.
If you want to see exactly what the AI is telling your buyers about your category right now, that is the first thing a paid audit measures, and the full method is on the methodology page.