Half of B2B software buyers now start their research with an AI chatbot. See the questions buyers ask at each stage and the content that answers each one.
By , founder and lead strategist at Flowup
Direct answer
The B2B buying journey now starts in a conversation with AI. According to G2, 51% of B2B software buyers start their research with an AI chatbot more often than with Google, and the vendor shortlist begins to take shape there, before any contact with sales. To make that shortlist, a website has to answer the questions of each stage: problem, solution, vendors, price, comparison and risk.
Sources verified on October 3, 2026.
On April 15, 2026, G2, the software review platform, released a survey of 1,076 B2B decision makers across North America, Europe, the Middle East, Africa and Asia-Pacific. Half of them start their research with an AI chatbot more often than with Google; in April 2025, the figure was 29%. And 69% said the chatbot's guidance led them to choose a different vendor from the one they had in mind.
In May, ChatGPT began turning brand names into links to the brands' websites. According to Profound, which tracks this traffic, the visits ChatGPT sends to B2B software companies grew by more than 200%. And, according to BrightEdge, ChatGPT accounted for 95.1% of the visits that AI assistants sent to websites in August 2026.
The studies come from outside Brazil, Flowup's home market, and that needs to be clear: we found no equivalent public research with Brazilian buyers. But Brazil is among the three largest ChatGPT markets in weekly users, according to the Brazilian business magazine Exame (August 4, 2026). This guide shows what the buyer asks AI at each stage, which content answers each question and where human validation still decides.
Where to start, depending on your role
The change is not in the number of stages in a purchase, but in the place where the first vendor shortlist takes shape. More and more, it comes out of a conversation with an AI assistant, before any contact with sales, and the studies of 2025 and 2026 point in the same direction.
| Study | Who was surveyed | What it shows |
|---|---|---|
| G2, The Answer Economy (April 15, 2026) | 1,076 software purchase decision makers, in March 2026 | 51% start their research with an AI chatbot more often than with Google. AI chatbots are the source that most influences the vendor shortlist, cited by 54%. And 64% often encounter inaccurate recommendations. |
| Gartner, survey of B2B buyers (May 20, 2026) | 645 B2B buyers, from August through September 2025 | 45% used generative AI, primarily to gather information on vendors and products. And 69% prefer to validate AI-generated insights with a sales rep. |
| 6sense, The B2B Buyer Experience Report for 2025 | Nearly 4,000 B2B buyers | 95% of purchases come from the Day One shortlist. According to 6sense's guide on AI in buyer research (November 25, 2025), 94% use LLMs (large language models) during the buying journey. |
| TrustRadius, 2026 B2B Buying Disconnect Report (July 15, 2026) | 1,862 technology buyers, in January 2026 | 63% used AI during their purchase, and 94% of those fact-check its responses at least some of the time. 83% shortlisted three or fewer products. |
| Semrush, survey on AI in B2B buying (July 8, 2026) | 622 professionals in the United States, in March and April 2026; the percentages are of the 519 who use AI tools for work | 61% use AI to compare vendors and 53% to ask for recommendations. After the answer, 71% visit the vendor's website. |
The numbers do not match one another, because the questions and the audiences change from one study to the next. The direction, though, is the same: AI has entered the first half of the purchase, where the shortlist takes shape, and people still decide at the end.
Our reading
If nearly all purchases come from the Day One shortlist, and that shortlist increasingly takes shape in a conversation with AI, being left out of the conversation is costly. The buyer may never reach the company's website, or its sales team, because the decision about who gets to compete has already been made.
The B2B buying journey has five moments when the buyer talks to AI: understanding the problem, learning about the solutions, building the vendor shortlist, checking price and alternatives, and assessing risk. At each one, the buyer asks different questions, and AI answers with what it finds, on the company's website or in other sources.
The Semrush survey shows the most common uses among those who use AI tools for work: 66% explore possible solutions, 61% compare vendors, 59% try to understand the problem or the category more deeply and 53% ask for recommendations. The table below arranges these questions in the order of the purchase, with examples from a company looking for a fleet management system.
| Moment | What the buyer asks | What AI needs to find | Content that answers it |
|---|---|---|---|
| Problem | "Why have our fleet maintenance costs gone up so much?" | A clear explanation of the problem, with causes and signs | Guides and articles that explain the problem with data |
| Solution | "What kinds of solutions are there for this? Do I need software or a consulting firm?" | Solution categories and when each one fits | Solution pages and comparisons between approaches |
| Vendors | "Which companies do this in Brazil for mid-sized trucking companies?" (an example from Brazil, Flowup's home market) | Who the company is, what it does, for whom and where it operates, stated the same way in every source | Homepage, service pages, about page, directory profiles and reviews |
| Price and alternatives | "How much does it cost? What's the difference between X and Y? What are the alternatives to X?" | Price ranges, what drives the price and honest differences | Pricing page, candid comparisons and alternatives pages |
| Risk and validation | "Is company X reliable? Does it integrate with our ERP? Who has used it?" | Proof that can be checked and stated limits | Case studies with numbers, technical documentation, reviews and a "who it is not for" section |
The table shows where the work is. The first two stages are the classic territory of top-of-funnel content. The last three decide who makes the shortlist and who drops off it, and they are the ones B2B websites tend to leave unanswered.
The questions of the final stages are the ones B2B websites answer least: how much it costs, how the company compares with the alternatives and who it is not right for. When the website stays silent, AI answers with whatever it finds elsewhere. In the G2 survey, 64% of buyers say they encounter inaccurate recommendations often, at least a few times a month.
Transparent pricing has been the number one item on technology buyers' wish list for four years running, from 2023 to 2026, according to the TrustRadius report. Publishing prices does not require a fixed price list: ranges, the factors that push the price up or down and the engagement formats already answer the question. That is the approach of Flowup's GEO and AEO consulting page: the Brazilian version shows indicative amounts and what defines each project, while the English version describes the formats and the factors and states that, for international projects, the price is defined in a proposal.
If the company does not compare itself, AI does the comparing for it, with whatever sources it has. A candid comparison page says where the company is better, where the alternative is better and who each one is right for. Comparing vendors is the second most cited use of AI in B2B buying in the Semrush survey, at 61%.
Saying who the solution is not for seems risky, but it helps AI recommend the company for the right case and spares the sales team conversations with no fit. A "who it is not for" section answers a question the buyer asks anyway, with or without the company in the conversation.
Example
A fleet management software company asks ChatGPT how it shows up for "alternatives to system X for a mid-sized trucking company." The answer cites three competitors and says the company serves only large fleets, which has not been true since last year. Its website has no pricing page and no comparison page, and it does not say who the product is for. The three missing pages are the starting point, along with correcting the information in the external sources that AI cites.
Since May 2026, ChatGPT has been turning brand names into links to companies' homepages. As a result, the homepage now receives the buyer who has just read an answer about the company, and it has to confirm in a few seconds what AI said.
According to Profound's study of May 19, 2026, the change began on May 7, 2026. Before it, about 4% of the visits coming from OpenAI landed on the homepage; after it, 24%. Over the same period, daily visits coming from ChatGPT grew 60% to 65% across the sites Profound monitors, and more than 200% for B2B software sites. The Semrush survey shows the other side: after the AI answer, 71% of those who use AI tools for work visit the vendor's website and 63% search for the company on Google.
What the first screen of the homepage needs to answer:
The homepage also needs to say the same thing the other sources say about the company. If the AI answer describes a service the homepage does not show, the buyer distrusts one of the two. Consistent structured data helps systems connect the brand to what it does, but it does not fix copy that says something else.
AI builds the shortlist, but it does not close the purchase on its own. The same studies that show the advance of AI show the buyer checking what it said against people, trials and proof.
With so many people deciding, each one asks AI different things: finance wants to know about cost, IT about integration and security, legal about the contract. The content has to serve the committee, not only the person who made the first contact. That is the role of case studies with numbers, technical documentation, the sales presentation and the business proposal, which keep circulating among the committee members after the meeting.
In technical B2B, this proof of competence before the first contact is the core of the work, as we show in the article on marketing for engineering and B2B companies.
Mapping the buyer's questions means listing what the buyer asks at each stage, checking what AI answers today and tying each question to a page of the website. The work starts with the people who talk to the customer, not with the keyword tool.
Turning this map into an editorial plan and bottom-of-funnel pages is the work of Flowup's Growth Content, which starts from the buyer's questions, stage by stage.
What is not yet known
The difference between ranking and being cited, which lies behind all of this, is the subject of our guide to SEO and GEO.
In short
The B2B buying journey is the path a company travels from noticing a problem to hiring a vendor. It usually goes through understanding the problem, learning about the solutions, building a vendor shortlist, comparing price and alternatives and assessing risk, and it involves several people. According to Forrester, a typical buying decision includes 13 internal stakeholders and 9 external influencers.
AI has become the starting point of the research and the place where the vendor shortlist begins. In G2's April 2026 survey, 51% of B2B software buyers start with an AI chatbot more often than with Google, and 69% chose a different vendor than planned because of the guidance they received. Validation with people continues in the final stages.
No. Google ranks pages for a search; ChatGPT assembles an answer from several sources and cites some of them. The fundamentals, such as clear content and authority, count for both, but a company can rank well and be left out of the answer, or be cited without being at the top of Google. That is why the two kinds of presence are measured separately.
It is worth publishing at least ranges and what drives the price. Transparent pricing has been the number one item on technology buyers' wish list for four years running, according to TrustRadius, and without that information AI answers with whatever it finds on other websites. When the price depends on scope, the page explains the factors and the engagement formats.
Yes, if the comparison is candid. Comparing vendors is one of the most common uses of AI in B2B buying: 61% of the professionals surveyed by Semrush who use AI tools for work do it. Without a comparison made by the company itself, AI uses those made by third parties. The page should say where the company is better, where the alternative is better and who each one is right for.
Yes, but later and with a shortlist already in hand. According to 6sense, 95% buy from a vendor that was already on the Day One shortlist. According to Gartner, 69% prefer to validate AI-generated insights with a sales rep. The role of sales becomes validating, reducing risk and answering what AI left open.
The studies cited surveyed buyers in North America, Europe and Asia-Pacific, and we found no equivalent public research with Brazilian buyers. Our reading is that the shift is reaching Brazil, Flowup's home market, which is among the three largest ChatGPT markets in weekly users, according to the Brazilian business magazine Exame. Even so, it is worth measuring with your own buyer before deciding.
Sources: G2, press release on The Answer Economy survey (April 15, 2026) and The Answer Economy report (2026); Profound, branded links in ChatGPT and referral traffic (May 19, 2026); BrightEdge, ChatGPT referral traffic in 2026 (September 24, 2026); Forrester, The State of Business Buying, 2026 (January 21, 2026); Gartner, survey on validating AI-generated insights with sales reps (May 20, 2026); 6sense, The B2B Buyer Experience Report for 2025 and guide on AI in buyer research (November 25, 2025); TrustRadius, 2026 B2B Buying Disconnect Report (July 15, 2026); Semrush, survey on AI in B2B buying (July 8, 2026); Exame, ChatGPT começa a exibir anúncios no Brasil, um dos três maiores mercados da OpenAI (August 4, 2026, in Portuguese). Accessed on October 3, 2026.
The vendor shortlist is being built before the first meeting, in a conversation your company does not see. Knowing what that conversation says, and which questions your website still does not answer, is the first step to getting into it. Ranking is not enough. Be the answer.
Growth Content
Find out which buying questions your website does not answer yet
Flowup's free diagnosis shows how search engines and AI assistants read your website. Growth Content turns the buyer's questions into pages that answer each stage of the purchase, with no promise of rankings.
Founder and lead strategist, Flowup Agency
Guto Bertoncini is the founder and lead strategist of Flowup Agency, which he has run since 2011. He is the author of the B.I.N.A. Method, Novo SEO and the Base Informacional Semântica (Semantic Information Base), and leads the agency's SEO for AI, GEO and AEO practice, preparing companies to be found on Google and cited by artificial intelligence platforms. He writes about search and AI on the Flowup blog and on his official website.
