Ask buyers in your sector how they found their last supplier and you will hear three different answers, and only one involves typing your service into Google. Some still work off a recommendation. A growing number describe the problem to an AI assistant and act on the names it gives back. How buyers find suppliers online has changed twice in thirty years, from the printed directory to the search engine, and the third shift is already underway. In the first four months of 2026, 68.01% of US Google searches ended without a click, according to SparkToro’s June 2026 analysis of Similarweb clickstream data. Here is what each shift did, and what this one does to your enquiries.
Key takeaways
- Business discovery has moved through three surfaces: the printed directory, the search engine, and now the AI answer.
- Each shift showed buyers fewer suppliers than the one before it. The market did not shrink, the shelf did.
- This shift is quieter than the last two because the loss happens off your website, where your analytics cannot see it.
- You can hold position one in Google and still not be named when a buyer asks an assistant the same question.
- It is checkable in about a minute, which is the only reason it is worth worrying about at all.
How do buyers find suppliers online now?
Buyers find suppliers online in three overlapping ways: a search engine, a directory or marketplace listing, and increasingly an AI assistant that returns a short list of named businesses rather than a page of links. All three run side by side. The difference is that the newest one hands the buyer a decision instead of a set of options.
Set the three eras next to each other and the pattern is hard to miss.
| Era | How buyers found you | What winning meant | Who got left behind |
|---|---|---|---|
| The printed directory, to the late 1990s | A classified listing under a heading, chosen from a column of competitors | Being listed under the right heading, with a bigger entry than the firm above you | Businesses that were not listed, or were listed under a heading nobody read |
| The search engine, roughly 2000 to 2024 | A query typed into a search box, answered with ten links and a map | Ranking on page one for the terms buyers actually typed, then converting the click | Businesses with no website, or a website search engines could not read properly |
| The AI answer, now | A problem described to an assistant, answered with a handful of named businesses | Being named in the answer and cited as a source, whether or not the buyer clicks | Businesses that are visible to a crawler but not legible to a model, with little a third party corroborates |
Nothing in the left-hand column has disappeared. Directories still exist. Search still sends the majority of enquiries for most UK SMEs. The eras stack rather than replace, which is exactly why the newest one is easy to dismiss.
Why did each shift make the market smaller?
Because each new surface displayed fewer suppliers than the one before it. A directory page listed dozens of firms in a column and the buyer did the filtering. A results page realistically gives you the top handful. An AI answer names three to five businesses, sometimes one.
That is the compression, and it is the whole story. Every shift in discovery has taken the shortlist out of the buyer’s hands and moved it earlier, into the surface itself.
The commercial consequence is that the gap between being on the shortlist and being nearly on it keeps getting more expensive. In the directory era, being fourth choice still got you seen. On an AI answer, there is no fourth choice to be.
Why is this shift quieter than the last two?
Because the loss happens off your website. When directories faded, the phone stopped ringing and you knew. When search took over, your traffic told you. This time your rankings can hold, your analytics can look steady, and the shortlist that excluded you was drawn in a conversation you were never shown.
There is no impression data for it. No query, no referrer, no bounce. Analytics measure the people who arrive, never the ones who never considered you, and that has never mattered as much as it does now.
The click-level evidence points the same way. Pew Research Center, publishing in July 2025 from the browsing data of 900 US adults, found that people clicked a traditional search result in 8% of visits where an AI summary appeared, against 15% of visits where one did not. Pew is careful to call that an association rather than a cause, and both that study and the SparkToro figure above are US data, so treat the magnitudes as directional.
The UK picture is thinner but moving the same way. Ofcom’s Online Nation report, published in December 2025, recorded 1.8 billion UK visits to ChatGPT in the first eight months of 2025, up from 368 million in the same period of 2024.
Why do businesses miss a shift like this?
Rarely because they cannot see it. Usually because their incentives point at the old surface. The numbers that get reported, the budgets that get approved and the targets people are measured on all describe the channel that used to work, so new behaviour looks like a rounding error for longer than it should.
MapQuest is the cleanest version of this. It was where people went for directions, it was very good at that, and its business depended on people going there. When directions stopped being a destination and became a feature inside search and inside the phone, the product was not beaten on quality. The surface moved and the incentive was to defend the old one.
That is the lesson worth taking, and the only one. Not that anybody was stupid, and not that your enquiries are about to collapse. They will not. They will thin out quietly instead, one quarter at a time, while every report you read says the rankings are fine.
How do you check whether it is happening to you?
Ask the assistants what your buyer would ask, in your buyer’s words, and read who gets named. Do it in ChatGPT, in Perplexity and in Google’s AI Overviews, because they frequently disagree with each other. Then compare those names with who ranks for the same service in Google.
- Write down the three questions a buyer asks before they shortlist anyone. Not your service names, their problem, in their phrasing.
- Put each question to two or three assistants, and note every business named and every source cited.
- Run the same questions through Google and compare the two lists side by side.
Where the lists do not match, you are ranked but not recommended, and that gap is the thing worth fixing.
One honest caveat. No platform publishes how it selects the businesses it names, and answers vary between sessions, accounts and phrasings. Treat what you find as a sample rather than a score, and never accept a promise of guaranteed citations from anyone, including us. What you can do is make your business easier to verify, easier to quote and better corroborated by sources outside your own website, which is the substance of AI search optimisation.
Who should be doing this work in your business?
Someone senior enough to change the plan when the answer comes back badly, not someone whose job is to keep the current plan running. The first two shifts in discovery rewarded businesses that acted while the behaviour was still a minority, and that is a judgement call rather than a task.
Lobo Digital’s founder, Marcus Lobow, spent seven years in-house across facilities management and property before starting the agency, and has over ten years in B2B marketing. He has sat on the client side of this exact decision, with a budget he had to defend internally and an agency report that did not answer the question he was asking.
The work that follows is unglamorous. Get the fundamentals right, answer buyer questions plainly, get corroborated elsewhere. Doing exactly that took one organic channel from 40 to over 1,000 monthly visits on commercial terms inside a single month. These results were achieved by Marcus Lobow, some in-house as a senior marketer and some through Lobo Digital.
Every account here is run by that same senior marketer. The person you brief is the person who does the work.
Frequently asked questions
Has search volume actually fallen?
No. Searching is up, clicking is down. SparkToro’s June 2026 analysis of Similarweb clickstream data found 68.01% of US Google searches ended without a click in the first four months of 2026, against 60.45% in 2024. People are still asking the questions. More of those questions are now being answered on the results page rather than on yours.
Is this only a consumer trend?
No, though the evidence is uneven by sector. Ofcom’s Online Nation report, published December 2025, recorded 1.8 billion UK visits to ChatGPT in the first eight months of 2025, up from 368 million a year earlier. In B2B the strongest published research covers software buyers specifically, so treat it as the best available proxy for business buying behaviour rather than a figure for your own sector.
What happened to the businesses that read the last shift early?
They were unremarkable about it. They built a proper website while competitors were still buying directory space, got the fundamentals right, and collected several years of cheap enquiries before the rest of the market arrived. Nothing dramatic happened to the ones who waited either. Their enquiries thinned out quarter by quarter, and the cost of winning each new one went up.
Does this apply to a business with no website traffic problem?
Yes, and those businesses are the most likely to miss it. Stable traffic tells you that the people who find you still arrive. It says nothing about the buyers who described their problem to an assistant, received three names and never reached a results page at all. That gap does not show up in analytics, which is why it has to be checked directly.
Find out where you stand
You can see the gap yourself in about a minute, for nothing, and keep the findings whether or not you ever speak to us.
Run the free AI visibility check
Questions about what the result means for your sector, email hello@lobodigital.co.uk and they come to a senior marketer, not a queue.
