How B2B buyers research software in AI search.

Your buyers build the vendor shortlist inside an AI chatbot before they ever talk to sales. The data on where they start, what they trust, and why "are we ranking" is now the wrong question.

The shortlist forms
before the call.
In 2026, B2B software buyers start their research inside an AI chatbot and build a vendor shortlist from the companies it names - often before they visit a website or talk to sales. G2's March 2026 survey of 1,076 B2B decision makers found 51% now begin software research with an AI chatbot more often than with Google, up from 29% in April 2025. If the answer does not name you, you are not on that buyer's shortlist.

I'm Gen Furukawa, founder of SuperMarketers. I build AI visibility systems for B2B SaaS companies and run AI search audits across dozens of them. The pattern is consistent: the buyer's first move is no longer a Google query and a list of blue links. It's a question typed into ChatGPT, and a short answer that names two or three companies. Everything below is sourced to named third-party research; where a number is ours, it is labeled as ours.

The reframe for a founder is simple. The real question is no longer "are we ranking?" It's "are we in the answer?"

The shift: from Google to answer engines

The surface where research begins has moved, and the move is recent enough to measure. In G2's March 2026 survey of 1,076 B2B decision makers, 51% said they begin software research with an AI chatbot more often than with Google, up from 29% in April 2025. A broader 71% said they rely on AI chatbots for software research, compared with 60% seven months earlier. That is not a slow drift. That is the starting line moving inside a single year.

Gartner's 2026 research found the same direction from a different angle: 45% of B2B buyers reported using generative AI, primarily to gather information on vendors and products. The scale of the surface is hard to argue with. OpenAI's Sam Altman said in October 2025 that ChatGPT had passed 800 million weekly active users. Your buyers are already there. The question is whether your company is in what the engine hands back.

Why this is different from SEO

SEO moved the buyer toward a link. An answer engine resolves the question and names a short set of companies. The buyer can finish the research step without ever clicking through. That changes what visibility means: not a ranking position, but whether the engine says your name. We unpack the budget tradeoff in SEO vs AEO for B2B SaaS.

Where buyers start the journey

When a buyer opens an AI chatbot to research software, ChatGPT is where most of them are. G2 found ChatGPT is the dominant chatbot for B2B software research at 63%. The rest of the research happens across Perplexity, Google AI Overviews, Claude, and Gemini, but the front door for the largest share is ChatGPT.

What happens next should reorganize your priorities. The buyer asks a category question: "best [category] tools for [use case]." The engine returns an answer that names a handful of companies and describes each one. The buyer treats that answer as the starting shortlist. They do not scroll through ten results and form their own opinion. They inherit the engine's opinion, then verify it.

MetricFigureSource
Start in AI chatbot more than Google51% (up from 29% in April 2025)G2, March 2026 (n=1,076)
Rely on AI chatbots for software research71% (up from 60% seven months prior)G2, March 2026
ChatGPT share of AI software research63%G2, March 2026
Used generative AI for vendor and product info45%Gartner, 2026
Chose a different vendor based on AI guidance69%G2, March 2026
Bought from a vendor they had not known before33%G2, March 2026

Read the last two rows again. Sixty-nine percent of buyers chose a different vendor than they originally planned to, based on AI chatbot guidance. One in three bought from a company they had never heard of before the engine named it. That is the upside if you are cited, and the cost if you are not. The shortlist is being rewritten inside the answer.

"The real question is no longer are we ranking. It's: are we in the answer?"

What this means for your shortlist

The shortlist forms early, and it forms with strong preferences already loaded. Forrester's 2024 Buyers' Journey Survey found that 92% of B2B buyers start the buying process with at least one vendor already in mind, and 41% have a single preferred vendor before formal evaluation begins. B2B buying is now closer to confirmation than discovery.

Put the two findings together. Buyers arrive at the formal process with a preferred vendor already chosen, and increasingly that preference is shaped inside an AI answer during the research phase. The contest is not the demo or the proposal. It is the moment the engine generates the shortlist. If you are named there, you enter the evaluation as a contender. If you are not, you are spending sales effort to overturn a preference that formed without you in the room.

This is also where "found" and "recommended" stop being the same thing. Your page can be perfectly indexed and crawlable - found - and the engine can still answer the buyer's question by naming three competitors. Recommended is the only state that puts you on the shortlist. The metric that captures it is your citation rate: the percentage of buyer queries where the AI recommends you by name.

The cost of being invisible

The pipeline math is unforgiving. A buyer who searches, gets an answer, and builds a shortlist that excludes you has just completed the most important step of their journey without ever knowing you exist. You will never see that as a lost deal, because it never became a deal. It became a competitor's deal.

Traditional search no longer rescues you here either. The Pew Research Center analyzed 68,879 Google searches in March 2025 and found that when an AI summary appeared, users clicked a traditional search result in just 8% of visits, versus 15% when no summary appeared. Only 1% clicked a source link inside the summary itself. The click you used to earn from ranking is increasingly resolved inside the answer. If your strategy depends on the buyer clicking through to your site, that path is narrowing.

From our audits

In our AI search audits across dozens of B2B SaaS companies, the most common finding is not a low ranking - it is total absence from the category answer. The company has a strong product and a large blog archive, and the engine still names competitors instead. The fix is an AI visibility system that tracks the same buyer queries every month and closes the gaps. We are publishing the aggregate benchmarks from our multi-engine scans in a forthcoming State of B2B SaaS AI Visibility report; until then we report only what we observe, never a market-wide number we cannot source.

What founders should do about it

You do not fix this with more blog posts. You fix it by becoming the company the engine names for your category's core questions. Four moves, in order.

  1. Baseline where you stand.Open ChatGPT, Perplexity, and Google AI Overviews. Run the 5 to 10 questions your buyers actually ask. Record who gets named and whether you appear at all. The full process is in our 30-minute AI visibility audit.
  2. Fix your entity description.Write one specific, unambiguous 40 to 60 word description of what your company does and who it is for. Use it verbatim on your homepage, your LinkedIn About, and anywhere you are described externally. Engines need a clear entity before they will name you.
  3. Rebuild your core pages for extraction.Direct answer in the first 100 words, numbered steps, self-contained FAQ, schema markup. Structure is what an engine can lift into an answer. The specifics are in how to get cited by ChatGPT.
  4. Track citation rate monthly.Re-run the same buyer queries every month and score whether you were named. This number, not posts published, is the metric that tells you whether your AI search presence is improving.

None of this requires a large team. It requires a system: the same buyer questions tracked every month, your core pages kept structured and current, and gaps closed as they appear. That is the work that moves you from absent to recommended.

Sources and methodology

Every external figure here is attributed to a named source. Buyer-behavior figures are from G2's March 2026 survey of 1,076 B2B decision makers across North America, EMEA, and APAC. The generative-AI adoption figure is from Gartner. The vendor-preference figures are from Forrester's 2024 Buyers' Journey Survey. The click-through figures are from the Pew Research Center's March 2025 analysis of 68,879 Google searches. The ChatGPT user figure is from OpenAI, reported in October 2025. Any observation labeled "from our audits" is our own first-party finding, not a market-wide statistic.

How B2B buyers research software in AI search FAQ

How are B2B buyers actually researching software in 2026?
They start in an AI chatbot. In G2's March 2026 survey of 1,076 B2B decision makers, 51% said they begin software research with an AI chatbot more often than with Google, up from 29% in April 2025, and 71% rely on AI chatbots for software research. Buyers ask the engine to name the best tools for their use case, read the generated answer, and build a shortlist from the companies the AI names - often before they visit a single vendor website or talk to sales.
Do buyers really use ChatGPT to find vendors?
Yes, and ChatGPT leads. G2 found ChatGPT is the dominant chatbot for B2B software research at 63%. Gartner separately reported that 45% of B2B buyers used generative AI, primarily to gather information on vendors and products. The practical effect: 69% of buyers in the G2 survey chose a different vendor than they originally planned based on AI guidance, and 33% bought from a vendor they had not previously heard of.
What percent of buyers start their research in AI search?
According to G2's March 2026 survey, 51% of B2B software buyers now begin software research with an AI chatbot more often than with Google, up from 29% in April 2025. A broader 71% rely on AI chatbots at some point in their research. The exact figure varies by source and by how the question is framed, so always cite the survey, the sample, and the date rather than a single floating number.
Our organic traffic dropped year over year. Is this an AI Overviews problem?
It may be part of it. The Pew Research Center analyzed 68,879 Google searches in March 2025 and found that when an AI summary appeared, users clicked a traditional search result in 8% of visits, versus 15% when no summary appeared - and only 1% clicked a source link inside the summary. So a query that used to send you a click now often resolves inside the answer. The defensive move is to be the company named inside that answer, not just a blue link below it.
If I am not cited in AI answers, am I out of the consideration set?
Effectively, for that query, yes. If a buyer asks an engine for the best tools in your category and the answer names three competitors and not you, you are not on the shortlist that buyer just built. Forrester found that 92% of B2B buyers start the process with at least one vendor already in mind and 41% have a single preferred vendor before formal evaluation. The shortlist forms early, and increasingly it forms inside an AI answer.
How is being "found" different from being "recommended" in AI search?
Found means your page exists and is crawlable. Recommended means the engine names your company in its answer to a buyer's question. SEO optimized for found - rank the page, earn the click. AI search rewards recommended - the engine resolves the question and names a short set of companies. You can be perfectly indexed and still never be named. The metric that captures this is citation rate: the percentage of buyer queries where the AI recommends you.

Related AEO guides for B2B SaaS founders

◆ ◆ ◆
See exactly where you're invisible.

Run the free Displacement Report.