Contributor Content What's This?

What AI Search Knows Before the Sales Team Does

How a buyer can rule out a vendor before ever paying them a visit.

By Matthew Kayser | Nov 20, 2026
Adobe Stock

A buyer can spend 20 minutes comparing vendors and show up in a website traffic report as someone who already knew the company’s name. They asked an AI search tool for a recommendation, read the answer, and looked up one of the businesses it mentioned. By the time the buyer calls, they might have a specific product in mind. The sales rep gets a name and a phone number, with no record of the conversation that got the buyer that far.

Vijay C. Jacob, who founded AEO Engine with Dan Ashburn, says he started hearing about that missing conversation on calls with prospective clients. He’d ask how they found the company, and some would tell him they asked a chatbot. 

That’s a useful answer to a question plenty of sales teams already ask, yet it leaves the second question for the person taking the call: what did the AI tell the buyer before they picked up the phone?

The word ‘portable’ leaves out a lot of context

A company buying a display for a trade show can spend an afternoon looking at things described as portable. That word can have different meanings depending on the company and what tools they have access to. 

Booths that ship in a crate can qualify for one company, but another may need to wheel the item from a taxi into the venue. The difference becomes important when the employee is working the show alone, and the crate is already at the curb.

That’s the sort of detail a buyer can put straight into a chat. They can ask how many people will be there to set up the booth and which models come in a case that a single person can lift. 

A vendor’s product page may list the packed weight, while its dealer may indicate the weight of the display on its own. Both pages can be right about weight, but if the figures describe different things, a buyer may be in for a rough day.

The person at the sales desk may be the first one at the company to hear about this kind of issue. A customer who asked whether the case is included might sound like someone who skimmed the product page. They could even be looking right at a comparison that’s already assured them the number is correct. There’s a difference between explaining a specification and finding out that the customer has been shopping with the wrong one.

A click comes late into the sequence

According to Pew Research Center, people in its March 2025 browsing study clicked a traditional result on 8% of Google search visits with an AI summary, compared to 15% of visits without one. A link in the summary itself drew a click on 1% of visits where a summary appeared. That’s a look at what people did on those search pages, but it doesn’t tell a business whether someone wrote down a name, came back to it later, or went with another vendor entirely.

That distinction is easy to lose when the traffic report comes up in a meeting. A company can have fewer visits and sales inquiries. In that case, the drop deserves about as much attention as it did before AI summaries existed. It can also have fewer visits and more inquiries about a particular product. 

The second business has something specific to check with its sales teams. Maybe people are arriving with a better idea of what they want, or a dealer sent over a batch of referrals the website report doesn’t explain.

Even a referral that a company can trace to a chat comes with some missing information. The buyer may have clicked to check the price, or they could have been trying to confirm whether the product existed at all. If the business wants to know which it was, there’s a good chance the person answering the phone can learn more in a minute than the one looking at a website traffic report can learn all afternoon.

What the podcast experiment could show

Jacob says his team tried its AI-assisted publishing process on its own material before using it on client sites. It launched  The AEO Engine AI Search Show and distributed the episodes and transcripts across several platforms. 

The team also published a comparison article that promoted the series. He notes that the podcast started appearing in answers to questions about shows in its field, though the team changed several things at once. A recommendation couldn’t tell them which one made the difference.

There’s a more immediate way to see what a transcript changes. If a guest makes a claim halfway through an episode, a person trying to check it can find the passage and read what came before. An editor can then see whether a caveat belongs in the same sentence as the claim. A producer who discovers a mistake after the episode goes out has somewhere to put a correction people can find without replaying the whole thing.

Jacob says his team leaves the choice of what to pursue to staff, uses software for parts of the production work, and brings in editors or engineers for the final review. In a search answer, a sentence may travel further than the paragraph it came from. The editor who moves a condition closer to the claim it qualifies is doing work that matters beyond the page where those words first appeared.

In a 2025 study, researchers at the Tow Center for Digital Journalism gave eight AI search tools excerpts from news stories and asked them to identify the articles. The tools returned wrong answers and sometimes linked to copies instead of the original publications. It was a test of news attribution, with its own limits. 

Its relevance to business stems from the familiar experience of clicking a link that looks perfectly convincing and finding out it supports something a little different from what the answer said. 

A dealer’s old product page can tell a customer a lot. The dealer may have copied the manufacturer’s warranty terms when it first started carrying a model. Years later, the manufacturer has revised the warranty, and the dealer’s page is still there with a product photo and a phone number. 

If that’s the page an AI answer pulls from, a customer can come away with a reasonable explanation for why they expect a repair the manufacturer says it doesn’t cover.

A website authority checker can give a business a score for the domain, which is useful for a different kind of comparison. The warranty problem lives on a particular page. Someone has to find the sentence, check it against the current terms, and ask the dealer to update it. That’s work the business can do even if it has no way to change the answer a customer already read.

Some businesses track an answer share, which is the proportion of sampled answers that name a company. A search for the best possible display and another for one that a person can carry solo might go into the same spreadsheet. They’re asking about different aspects of the product, though. A team that combines them into one figure can miss the reason its display kept appearing in one set of answers and disappearing from the other.

There’s also a distinction between getting named and recommended. An answer that says a booth is too heavy for one person may have mentioned the brand, and it might have done the buyer a favor. If the manufacturer makes a lighter model, the useful thing to check is whether the answer took one model’s weight as a fact about the whole line. That gives the company something to investigate beyond whether it’s moved up or down in a tally.

The record can become more useful when it includes the exact question, date, and the pages the answer pointed to. A team that sees the wrong weight has a source to inspect. If they notice the same figure again and the buyer brings it up on a call, the sales rep and the person who maintains the website finally have something concrete to compare. Both can find out where the number came from before deciding what to change.

Back on the sales call

There’s a limit to what a company can learn by running its own searches. Staff know the product names and understand how the company describes the differences between models. A buyer may come at the purchase with a completely different understanding. The phrases the website team has been optimizing might never occur to someone whose real concern is getting a display into a hotel ballroom before the loading dock opens.

That’s why the conversation Jacob describes may be more revealing than a screenshot of the recommended company. A buyer who remembers what they asked can tell the sales team about how they found the company, but also about the job the buyer needs done.

Even the person who remembers only half of the answer might recall the detail that got them to make the call. Sometimes that small observation is the one the company would have buried in a PDF.

The buyer at the curb probably won’t remember every name the search tool offered. They’ll recall whether the display came in a case they could get through the door on their own. If they had to phone the sales desk to find that out, the business has learned something worth putting on the product page, and the next person may find it in a chat or an ordinary search.

Either way, it’s a helpful piece of information to send out into the world alongside the word ‘portable.’

Entrepreneur Media was not involved in the creation of this content.