This guide explains how buying has changed now that buyers ask AI, and why that matters to a business that sells to them. It is written for owners, executives and marketing leads who want to understand the shift before deciding what, if anything, to do about it. It is not a pitch: most of it applies whether or not you ever work with us.
How buyers discover businesses now
For twenty years, discovery meant search. A buyer typed a few words, scanned a list of links, opened several tabs and worked out the answer themselves. A business could see that journey, at least in part: the rankings, the clicks, the visits, the form fills.
Now a growing share of buyers skip the list. They describe their situation to an AI assistant and ask it to do the legwork. The assistant explains the options, names a handful of businesses, compares them and suggests what to do next. The buyer may then visit one site, or none.
The evidence for this comes from several directions:
- In G2's March 2026 survey of 1,076 B2B software buyers, 51% said they start their research with an AI chatbot more often than with Google, and 54% ranked AI chatbots as the biggest influence on which vendors made their shortlist.
- In Alchemer's August 2026 survey of 1,002 US shoppers, 48.5% had used AI to research a purchase.
- In the first four months of 2026, 68% of US Google searches ended without a click, according to a study reported by Search Engine Land, and AI answers are a likely contributor.
None of this means search has gone. It means the first conversation about a purchase increasingly happens somewhere most businesses have not yet looked.
From traditional search to AI-assisted discovery
The difference is not only the interface. Three things changed at once.
| Search | AI-assisted discovery | |
|---|---|---|
| What the buyer types | A query aimed at a list | A question that describes their situation |
| What comes back | Ten links to read and compare | One written answer, often with a shortlist |
| Who does the comparing | The buyer | The AI, before the buyer sees anything |
| What a business can see | Rankings, clicks and visits | Usually nothing until a buyer arrives, if they do |
| What decides inclusion | How well a page ranks | What the AI knows, what it can read, and how it judges the fit |
A page can rank first in search and still be left out of the AI's answer, or described wrongly in it. That is why a business's search reports can look healthy while its position in AI is weak, and why the usual tools do not show the problem.
How buyers use AI to research products and services
Buyers do not use AI the way they used search. Five habits stand out.
The assistant supplies the vocabulary. Search rewarded buyers who already knew the term to type. With an assistant, a buyer can start from a symptom, such as a sealant that keeps failing in the cold, and be handed the category name, the standards that apply and a list of suppliers in one reply. A business whose pages only speak its own category language ends up being introduced by whoever explains the symptom best.
They bring their context. A buyer adds their budget, location, size, constraints and history, and expects an answer fitted to them. The more specific the situation, the more the answer can favour a specialist over a generalist.
Doubt turns into checking, not into searching wider. Buyers are wary of what AI tells them: only 35.4% of shoppers in Alchemer's survey trusted its recommendations mostly or completely, and Gartner found 69% of B2B buyers prefer to validate AI insights with a sales rep. Yet that checking stays inside the list they were handed. When an AI answer leaves a brand out, G2 found just 24% look for peer feedback; 21% accept the AI's pick and 14% fall back on a brand they already know. For a business that is not on the list, the buyer's caution does not help.
Buying groups multiply the conversations. Forrester's State of Business Buying 2026 puts the average B2B purchase at 13 internal and 9 external people, and roughly double that where generative AI features are involved. Each person may brief a different engine in a different way, so one business can be described several ways before the first meeting.
They enter at any stage. A buyer can arrive with a vague problem, a shortlist to check, or a choice already made, and pass through every stage of awareness inside one conversation.
We call the context of any question a buyer moment. It has six parts, and changing any one of them can change the answer.
A buyer who is a decision-maker, just looking, asks ChatGPT in Dutch, in the Netherlands with a short question at the opening question.
Illustrative. Click a part to change it.
How AI creates recommendations and shortlists
Engines draw on two kinds of knowledge, and the balance between them explains much of what businesses see.
- What the model already holds. Training leaves each model with a compressed, dated picture of the web. Well-documented companies can surface from that picture alone, usually without a source attached; newer or smaller ones mostly cannot, and changes to a business only reach it at the next retraining.
- What it fetches in the moment. For current or specific prompts, the engine searches and builds its answer from the pages it reads, often with links. Google calls its approach "query fan-out": one prompt is split into many related searches. This is where a business can move fastest, because the pages being read are pages it can improve.
How far each engine leans on one or the other varies, and so does its readiness to name companies at all. In our study of 240 AI conversations about money, a buyer who only asked "what is my first step?" was shown a firm in 1 of 24 conversations on ChatGPT, 6 on Claude and 16 on Gemini. Same buyer, same question, three very different markets.
Answers also vary from run to run. SparkToro had 600 volunteers run the same prompts nearly 3,000 times and found less than a 1 in 100 chance of the same list of brands appearing twice. In our own study, about 40% of the brands named appeared once and never again. So a single answer is a sample, not a verdict. A business's real position is the share of answers it appears in, per engine.
The shortlist itself is shaped by more than the business's own website: directories, review sites, forums, comparison pages, public registers and news coverage all feed the answer. Often the source that decides how a business is described is one it has never looked at.
How AI compares businesses and alternatives
Once a buyer has names, they ask the AI to compare them: "X or Y for my situation?", "Is X any good?", "What are the alternatives to X?". At this stage, being understood matters more than being found.
Comparisons go wrong in a few recurring ways:
- A fact is missing. The buyer needs a price, a certification, a lead time or a coverage area, and the engine cannot find it, often because it sits in a PDF, an image or behind a form. A rival whose facts are readable looks more complete.
- A fact is wrong or out of date. An old listing, a past price or a retired product is repeated with confidence.
- The framing is someone else's. The engine adopts a competitor's comparison page, or the most-cited source's view of the category, and the business is framed as a poorer fit for the buyer's situation than it is.
- The wrong company is meant. A similar name, or a missing link between a business and its products, leads the engine to describe someone else.
Engines also send buyers to check different things. In our study, the sources each engine linked for the same questions were noticeably different, so the proof a buyer sees depends on which app they opened.
How AI influences purchasing decisions
AI does not only list options. It frames the decision.
It can tell a buyer that a category is not mature enough yet, and talk them out of a purchase altogether. It can steer a buyer to a textbook answer, the default solution its sources mention most, which makes it easy for specialists and newcomers to be framed out. It can anchor a buyer on a figure: in our study, the same engine put one couple's maximum mortgage anywhere from €400,000 to €550,000 across ten runs, and the buyer arrives believing whichever number they saw.
And it changes who wins. In G2's survey, 69% of B2B software buyers said a chatbot led them to a different vendor than they had planned, and a third chose one they had not been considering. A business that was never on the buyer's list can be put there, and a business that was can be taken off it.
At the end of the conversation, the AI points the buyer somewhere: a website, a marketplace, a booking platform, a phone number. A business can be chosen and still miss the sale here, to a broken contact path, a marketplace that takes the margin, or a rival the AI found easier to reach.
Why businesses need to understand how AI represents them
Every business now has a version of itself inside AI, one per engine, assembled from whatever got written down about it. Some of it is current. Some of it is out of date with the date missing. All of it is repeated to every buyer who asks.
It comes down to three gates, and each is an opportunity as much as a risk:
Is the brand among those AI deems fit for the buyer's context?
Are its offerings, compliances, strengths and differentiators presented as the brand would?
Can the buyer, or an agent on their behalf, book, request a quote or buy?
| Gate | The question | When it works | When it does not |
|---|---|---|---|
| Found | Does AI name the business at all? | A place on the shortlist, where the decision now starts. | Buyers shortlist without it. |
| Understood | Does AI describe it correctly, for this buyer? | Buyers arrive understanding how it fits their situation, with their questions already answered. | Wrong facts, a weaker price position, or being framed as a poorer fit for their situation. This can cost more than absence, because the business falls out of the buyer's AI journey before they reach it. |
| Actionable | When the buyer is ready, can they act, with this business? | The ready buyer lands where they can act, with it. | Ready buyers end up with a competitor or a marketplace, or their agent abandons the task on its site. |
These gates multiply. A business that is found but misdescribed, or understood but hard to reach, keeps only part of the value of the gates it passes. So the weakest gate is usually where the most is to be gained.
The good news is that the causes are usually ordinary and fixable: a fact that is not stated in plain text, a stale listing, a page that does not answer the question buyers actually ask, a crawler that is blocked by a default setting. You cannot edit an AI's answer. You can change what is behind it.
The emergence of AI agents and agentic commerce
Search left the comparing to the buyer. AI assistants took the comparing over. Agents are starting to take over the doing as well: sending the quote request, booking the slot, placing the order. At every step a business gets fewer chances to make its case to a person, so what a machine can read about it carries more weight.
Signs of this are already visible:
- Google's information agents keep track of what a buyer wants and tell them when a match appears.
- In retail, Google's Universal Commerce Protocol lets agents inside AI Mode and the Gemini app read live prices and stock and put items in a basket.
- General-purpose browser agents can operate ordinary websites today, but they are brittle: a changed layout or a bot check is often enough to stop them.
For an agent, each gate asks something slightly different:
- Found: the agent's shortlist is limited to businesses it can discover and parse. Brand recall does not help it.
- Understood: the agent weighs specifications, prices, terms and availability. Claims it cannot verify count for little.
- Actionable: the deciding gate. The agent needs a route it can complete without a person: a quote request, a booking or an order, and afterwards an order status or a return.
The ground is still settling, and there is no clear winner among the standards yet. For most businesses the right move is a timed decision, not a rush: build now, wait, or use a provider who already handles it.
What businesses should do to prepare
You can start without anyone's help.
- Write down how your buyer describes their problem without knowing your name. Use their words, not yours.
- Ask it on ChatGPT, Gemini and Claude, then add the details a real buyer would. Ask "what is my first step?" in one conversation and "who should I go to?" in another.
- Repeat it. Ask each question several times on each engine, and in each language you sell in. Note which names come up, how you are described, and which sources are linked.
- Check your basics. Are AI crawlers allowed on the pages you want read? Are your price, coverage, certifications and contact details in plain text on a normal page?
- Ask your next three new customers what an AI told them before they contacted you.
That will tell you more about your buyer journey than most dashboards. What it will not tell you is why the answers are what they are, which of the causes matter for your business, and what to change first. That takes a structured approach: your buyers' real questions, run on every major engine many times, read at all three gates, with each finding traced to its source. We call that discipline AI buyer intelligence, and the way we do it is The Check.
Where to start
If you would like to see this for your own business, start with the free demo. Send us your URL, and a buyer or prompt you care about if you have one. We show you the buyer AI imagines for your business and that buyer's conversation with AI: whether you were named, how you were described, and who was named instead. If it is worth going further, we say so. If it is not, we say that too.
Sources
- G2, "The Answer Economy: G2's 2026 AI Search Insight Report," March 2026, n=1,076 B2B software buyers. https://learn.g2.com/g2-2026-ai-search-insight-report
- Alchemer, "Half of shoppers use AI to shop. Only a third trust what it tells them," August 2026, n=1,002 US shoppers. https://www.alchemer.com/content/industry/retail/ai-shopping-trust-gap-2026/
- Gartner B2B buyer survey, May 2026, via Demand Gen Report. https://www.demandgenreport.com/industry-news/news-brief/gartner-ai-is-reshaping-b2b-buying-but-human-sellers-still-close-the-confidence-gap/53046/
- Forrester, The State of Business Buying, 2026, via BIIA. https://www.biia.com/forrester-b2b-buying-groups-expand-as-they-question-ai/
- Search Engine Land, "Google zero-click searches hit 68% in early 2026." https://searchengineland.com/google-zero-click-searches-2026-study-479717
- Google Search Central, "AI features and your website." https://developers.google.com/search/docs/appearance/ai-features
- SparkToro, "AIs are highly inconsistent when recommending brands or products," 28 January 2026. https://sparktoro.com/blog/new-research-ais-are-highly-inconsistent-when-recommending-brands-or-products-marketers-should-take-care-when-tracking-ai-visibility/
- Google, "Universal Commerce Protocol updates improve AI shopping for retailers," 19 March 2026. https://blog.google/products-and-platforms/products/shopping/ucp-updates/
- CheckMarque AI, Three AI engines, eight Dutch money questions, 10 October 2026; How buyers decide inside an AI conversation and How AI engines decide which companies to name, 31 October 2026.
The complete guide to AI buyer intelligence: how AI discovers, represents and recommends businesses
What AI buyer intelligence is, what it measures, how to assess it, and how it differs from SEO, GEO, AEO and AI visibility tools. The reference guide to the discipline.
Read →Guide · · 10 minThe Check: a complete guide to the AI Buyer Audit
What The Check is, what it examines, how it works, what you receive and what happens after. The plain guide to CheckMarque AI's AI Buyer Audit.
Read →