admin 
I am shopping for a truck, and last week I walked into a dealership knowing more about the vehicle on the lot than the person selling it did.
That is not a claim about my research habits. I gave an AI assistant the VIN and asked it to tell me what I was looking at. Within a few minutes, it had confirmed what the truck actually was, which matters because listings mislabel trims. It had reconstructed the truck’s market history across every site it had ever been listed on, so an asking price turned into “this vehicle has sat for 280 days and has already dropped $5,000.” It had pulled the ownership and accident record, flagged the known problems for that model year, worked out how much factory warranty was left, and checked whether an out-of-state truck could be registered where I live. Then it wrote the questions I should ask, and none of them was “is this a good truck.” They were closer to “was that transmission replaced or just reprogrammed,” and “why has this been for sale for nine months.”
Every source it used existed 10 years ago. The government databases, the history reports, the owner forums, the manufacturer bulletins, the pricing aggregators. A determined buyer could have assembled the same picture over a week of evenings. What changed is that the assembly now takes minutes and requires no expertise at all. The information asymmetry between a buyer and a dealer, which the entire retail car business was built around, collapsed while I finished a coffee.
Last week I wrote about what happens to your judgment when the research gets compressed into seconds. Then I personally walked into a dealership feeling like I had done a month’s worth of homework. I could not have told you which parts of it I had actually verified. I was exactly the confidently underinformed buyer I described in that piece, and I was still better prepared for that dealership than the dealership was for me.
That is this whole article in one line. AI changed the buyer, and the business has not adapted yet.
Nobody did this to dealers. It is not a plot; it is a pace difference, and it is not specific to trucks. Every company that sells anything a person thinks about before buying is now facing the buyer I was last week. What follows is where the gap shows up, what the business is still doing in each of those places, and why the gap keeps growing without anyone deciding it should.
One disclosure, then I’ll get on with it. I sell a data measurement platform for exactly this problem, so this article is not going to tell you which tool to buy. It is going to tell you what the problem is, and why it is bigger than a tool.
You Are No Longer The First Conversation
The buyer arrives in the middle of a decision, not at the start of one. The research, the comparison, and the narrowing all happened in a chat window before your site loaded or your rep picked up. Pew Research Center surveyed 5,119 U.S. adults in February and found that about half now use AI chatbots, up from a third in 2024, that roughly a quarter use them daily, and that searching for information is the single most common thing people do with them. Pew notes the 2024 question was asked differently, so treat the jump as direction rather than a precise delta. A large share of your prospects had a prior conversation about your category, and you were not in it.
What the business still does is open as though it were the first conversation. The landing page explains what the category is. The discovery call walks through what the product does. The nurture sequence starts with the 101 email. All of it was built for a person standing at the top of a staircase, and last week’s piece covered why that staircase no longer starts at your front door. The mismatch is not that the content is wrong. It is that the content is addressed to someone who already left.
The question underneath this is what you say to a person who has already been told what you are. That is a positioning question with a timing component, not a content-volume question, and it is answerable once someone decides it is the question.
The Shortlist Got Written Somewhere Else
Ten results and a long tail became three names and a paragraph. That is what happened when the assistant compared the truck I selected against the whole market instead of against the listing. “Is this price fair” turned into “is this the best use of the money,” and four vehicles from three manufacturers ended up in a single comparison that none of them controlled.
For an established company this is the shift that stings, because the things it has spent years building do not count the way they used to. Fifteen years of customers, a working search footprint, a name people in the industry recognize. None of that is the evidence a model uses to decide whether a name belongs in a category answer. I have argued that the part of AI authority that gets you named takes years, not campaigns, and comes mostly from other people, and that when a model has nothing on your company it describes someone else in your place. Both of those are about the shortlist, and the companies asking “why not us” usually cannot say what is being said instead.
What the business still does is run positioning built to win ten slots. Competitive pages against the one rival it thinks about. Brand work measured by share of search. The shortlist is not ten long, and it does not care which rival you had in mind. The question that would actually move things is what the system treats as evidence of belonging, and that question has an answer. It is rarely asked because it does not clearly map onto anyone’s existing job.
Breadth Is Being Read, Not Visited
Every business that grew by having a page for every combination, product by use case, service by city, course by technology, was built for a system that listed pages. The assistant read a decade of forum threads, bulletins, and spec sheets about my truck and gave me back two paragraphs. The pages did their job. They got no visit for it.
The pages still matter, but for a different reason than the one they were built for. Being absent from what the models read is worse than being absorbed by it, and I have made that case before. The trouble is that the funding logic for the catalog is still “these pages rank and bring traffic,” and that logic is measurable, which is exactly why it never gets questioned. A cost center that reports cleanly does not get reviewed.
So the question becomes what the catalog is for now. If the honest answer is “to be read by a system and returned as a summary that may or may not name us,” then its structure, its consistency, and whether it agrees with what other sources say all matter more than its page count. That is a different production brief than most content teams have ever received, and nobody has handed it to them, because the old brief still looks like it is working.
The Credit Goes Nowhere
This is the one that generates the inbounds. A company discovers it is cited across AI Overviews, ChatGPT, Perplexity, and Copilot, sees almost no click-through from any of it, and asks whether that is fixable.
Common instrumentation has improved this year, and that’s a good thing for everyone, if you know what the data does and doesn’t tell you. Google added an AI Assistant channel to GA4 in May, and its own description says the feature tracks user clicks arriving from assistants. Microsoft shipped an AI Performance report in Bing Webmaster Tools in February that shows how often your pages are displayed as sources across Copilot and Bing’s AI answers. Both are useful, and I spent years building the second kind of tool, so I know what they can show and what they cannot. What they measure is the tail of a decision, a click that happened or a citation that was displayed. The influence on the buyer happened earlier, in a place where no click occurs, and no referrer-based instrument can record something that never produced a referrer. That is not a failing of the tools, it’s just the shape of the problem.
Last week, I argued that citation is presence, not traffic. The organizational version of that argument is different. Reporting, targets, and incentives are still built at the click, so marketing gets credit for what it can count, and the channel doing the most work on the buyer is the one that shows up smallest in the deck. Whether the clicks come back is the wrong question. If you are waiting for assistant referrals to grow until they rival what Google sends you, stop waiting. That is not how these systems work. An answer engine exists to resolve the question, not to refer it, and once the answer lands, the sources beside it stop being interesting, which is exactly the behavior the research measured last week. The numbers to watch are different ones: the share of questions in your category that now get answered by an assistant instead of a search result, and the total volume of clicks leaving Google’s results, not just yours. Both are moving in directions that do not reward a wait-and-see posture, and neither shows up in a report built at the click. What the answer says about you, and whether it is true, is the question that remains, and most companies have no reliable way to measure it yet.
This is the conversation I have with companies most weeks now. The first question is always which tool, and the tool market has problems of its own. The useful first question is what is being said about us, to whom, in which answers, and how much of it is accurate. Tooling comes third, after you know what you are trying to move.
Why The Gap Widens On Its Own
Each of those four places just discussed has the same thing underneath it. The buyer adapted at the pace of a software update, but the business adapts at the pace of a planning cycle.
The Pew numbers put a clock on it. Chatbot use went from a third of US adults to half in roughly eighteen months between surveys, and ChatGPT alone went from 18% of adults in 2023 to 44% in 2026. In that same window, most companies completed two annual plans. The change on the buyer’s side required no decision from anyone. People opened an app that was already on their phone. The change on the business side requires a budget line, an owner, a measurement, and a story for the board, and each of those takes a cycle to secure.
That is the entire asymmetry. “We are behind” is not a judgment about anyone’s marketing. It’s basic math. The buyer moves in weeks and months while the business moves in years.
The Window, And What Goes First
Three things I have written about separately are all, at their core, about time. The authority that gets a name into an answer accrues over years and comes from other people. When a model has nothing on a company, it fills the space with someone else (real or invented). And the correction that used to happen for free, when a curious reader kept clicking until they found the right version, now has to be paid for and fires slowly if it fires at all.
Put those together, and the window is not “act before the market notices.” It is that the shortlist for your category is being written now, from what is citable now, and the record hardens as it goes. The cost of being absent, or being described wrongly, rises every quarter you wait, not because competitors are racing you but because the substitute that got written in your place is getting cited, and cited things get cited again.
So here is the sequence, and I will keep it at the level of order rather than instruction, because the instructions are different for every company and the order is not.
- Know what is being said about you before you decide what to say.
- Decide what belongs in the answer before you decide where to publish it.
- Decide what to measure last, once you know what you are trying to move.
Companies do this in reverse. Tool first, metrics second, and causes, sometimes, third. Reversing the order is not a tactic, but it might be the difference between spending a year on this and spending a quarter.
AI changed the buyer in about 18 months. Businesses run on twelve-month plans. The gap between those two numbers is not a failure of anyone’s marketing. It is a timing problem, and timing problems are the kind that reward whoever starts first.
If you are seeing this in your own inbound, or in the gap between what your reporting says and what your prospects already know, tell me in the comments. I read all of them, and I am especially interested in anyone who has tried to answer the “what is being said about us” question before buying a tool to count it.
More Resources:
This post was originally published on Duane Forrester Decodes.
Featured Image: insta_photos/Shutterstock
