AI chatbots now sit at the start of B2B buying. A buyer describes the problem to ChatGPT, Gemini, Perplexity or Copilot, asks which vendors fit, and leaves with a shortlist before anyone on your team knows the deal exists. If you are not on that list, you may never hear about the deal.
This is already measurable. In G2's March 2026 survey of 1,076 B2B software buyers and decision-makers, 51% said they now start software research with an AI chatbot more often than with Google, up from 29% in April 2025.
This guide covers what buyers do in those chats, what the models read, what gets a vendor named, how the first sales call changes, and a checklist.
How B2B buyers use AI chatbots to research vendors
Buyers use chatbots to compare vendors, not just to learn about a category. G2 found that comparing vendor strengths and weaknesses is the most common use of AI chatbots in software research, named by 41% of buyers. Its full report shows two-thirds open with a category or competitor prompt: 33% start with a category prompt, 31% with a competitor prompt.
The chat sits next to search rather than replacing it: 71% of buyers rely on AI chatbots for software research, yet 80% still use Google somewhere in their buying journey. Gartner's survey of 645 B2B buyers, run in August and September 2025, found that buyers used an average of seven information sources in a recent purchase and that 45% used generative AI, mainly to gather information on vendors and products. The samples differ, the direction does not.
A typical session, as a made-up example:
- A category question. "Which companies can connect our ERP to a new warehouse system?"
- A comparison. "Vendor A or vendor B for a 200-person logistics company?"
- A price check. "What does managed security cost per month for 300 devices?"
- A shortlist request. "Three vendors to invite to a demo, with a reason for each."
The names in those answers are the shortlist, and shortlists are short: in TrustRadius's 2026 survey of 1,862 technology buyers, 83% shortlisted three or fewer products. Everything that follows, from the demo request to the RFP, starts from them.
How much do AI chatbots influence B2B shortlists?
More than any other source in G2's data. Asked what influences which vendors make their shortlist, 54% of buyers named AI chatbots, ahead of software review sites at 43%, vendor websites at 36% and vendor salespeople at 19%.
It changes real purchases: 69% of buyers chose a different vendor than planned because of AI chatbot guidance, one in three bought from a vendor they were not familiar with, and 85% think more highly of a vendor a chatbot recommends.
For a known vendor, that means deals lost in chats your sales team never sees. For a smaller company, it means a place on lists next to far bigger names, if what the model reads makes a clear case. How to tell if ChatGPT recommends your company shows how to test where you stand.
What do ChatGPT, Gemini and other AI chatbots read about your company?
Whatever they may reach and can read as plain text: your site, what others publish about you and the listings that describe you. Each assistant documents its search crawler; block it and you drop out of its answers.
Google's guide to AI features and your website says there are no additional requirements to appear in AI Overviews or AI Mode: a page must be indexed and eligible to show with a snippet. Its basics include allowing crawling in robots.txt and by your CDN or hosting, keeping important content in text form and making structured data match the visible page. Whether Google can index your key pages is the first check in our guide to auditing SEO on a B2B IT website.
OpenAI's crawler documentation says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers; GPTBot, which collects content that may be used for model training, is a separate setting. Perplexity documents PerplexityBot the same way and notes that firewalls may need an explicit rule for its bots. For Copilot, the AI Performance report in Bing Webmaster Tools shows how often your pages are cited in Microsoft Copilot and AI summaries in Bing.
Access is the minimum. What the model says about you comes from four places, and most companies control only the first:
- Your own pages: pricing, comparisons, alternatives, integrations, industry pages and case studies.
- Review platforms: G2 or Capterra for software, Clutch for IT services.
- Other people's content: comparison posts, industry roundups, analyst notes and forum threads.
- Profiles and listings: LinkedIn, business registers, partner marketplaces and directories.
How to get on AI shortlists: what gets a vendor named
Vendors get named when their pages answer the buyer's exact question with specifics, independent sources back them up, and their facts agree everywhere.
The GEO paper by researchers from Princeton University and IIT Delhi, accepted at KDD 2024, found that adding citations, quotations and statistics raised a source's visibility in generative answers by 30 to 40% on one of its two measures, while keyword stuffing performed poorly and a more authoritative tone brought no significant gain. We explain the study in what generative engine optimization actually is.
Proof from others counts too. In G2's survey, 45% of buyers said citations from software review sites are the most confidence-inspiring signal in an AI-generated answer. G2 runs review sites, so weigh that accordingly; still, a claim customers confirm elsewhere is easier for a model to repeat and for a buyer to believe.
So does consistency. G2's report says buyers see it as a trust signal when different chatbots describe a vendor the same way, and a red flag when they do not. If your site says mid-market banks, LinkedIn says startups and an old directory lists a retired product, the models have three stories to choose from.
The vendors that keep getting named usually have:
- Specific claims. Industries, integrations, rollout times and contract terms, written as facts instead of adjectives.
- A page for every buying question. Pricing, named comparisons, integrations, security and industries, each answering in its first lines.
- Published prices or ranges. G2's report says buyers use chatbots to work through pricing and packaging options, and TrustRadius reports that transparent pricing has been buyers' top wish for vendors four years running. With only contact us on your site, your price comes from somewhere else, or nowhere.
- Recent reviews where buyers look, requested as a routine step after go-live.
- Clear structure. Microsoft's Bing team writes that clear headings, tables and FAQ sections make content easier for AI systems to reference accurately.
How the first sales call changes when buyers bring an AI shortlist
The first call stops being discovery and becomes validation. The buyer already has a shortlist and a summary of each vendor, and wants to check it.
Gartner's findings, published in May 2026, say this directly: 69% of B2B buyers prefer to validate AI-generated insights with sales reps, even though 67% prefer a buying experience without a rep. Trust is split: 51% say they are more likely to meet misleading information from generative AI, 49% from a sales rep. Gartner concludes that sellers are moving from primary source of information to a source of validation and confidence.
Buyers know the tools make mistakes: in G2's survey, 64% run into inaccurate AI chatbot recommendations often or very often, and 24% turn to peer reviews when a chatbot contradicts a brand they trust. G2 advises sales teams to assume the buyer has already seen an AI-generated comparison with your competitors.
So change the first call:
- Ask what they already read. Which assistants, what did they say about us, who else is on the list? You learn the criteria you are judged on.
- Correct errors with evidence. A missing integration you have supported for a year? Show the documentation. A wrong price? Show your pricing page.
- Test the requirements the model assumed. Chatbots fill gaps with generic criteria. Ask which really matter to this buyer.
- Bring an honest comparison. One page against the two names the buyer mentioned, including where they are stronger.
- Log what the AI told them. A CRM field for AI claims, reviewed with marketing every month, becomes your list of pages to fix.
Validation does not remove the deal risk: a buying group can agree on an AI-built shortlist and still freeze at the decision, as why B2B deals stall in no decision explains. Our sales strategy development work builds these questions into your playbook.
Does outbound still work when AI builds the shortlist?
Yes, and most of all for companies the models do not name yet: if the shortlist comes out of a chat, an unnamed vendor gets in only by reaching the buyer directly.
Two things change. A prospect who gets your email can ask an assistant about you in seconds, so the model's view of you sits between your message and the meeting. And buyers no longer need a rep to explain the category: in G2's data, vendor salespeople influence shortlists for only 19% of buyers.
What works is a specific reason to talk: a trigger at the account, a fact the prospect may lack, an offer to help check their options. Our lead generation service runs that motion for you.
AI shortlist checklist for B2B companies
Work through it with marketing and one salesperson. Site fixes are quick; reviews and third-party mentions take longest, so start those first.
| Area | What to check | Fix |
|---|---|---|
| Crawler access | robots.txt, CDN and firewall rules for Googlebot, Bingbot, OAI-SearchBot and PerplexityBot | Allow search crawlers; decide on training crawlers such as GPTBot separately |
| Buying-question pages | Pricing, comparison, alternatives, integration and industry pages in plain text | Write the missing pages and answer in the first lines |
| Proof | Figures, named sources and customer facts on key pages | Swap adjectives for numbers with a source |
| Reviews | Recent reviews on G2, Capterra or Clutch | Ask every customer after go-live |
| Company facts | Same name, description, markets and prices on your site, LinkedIn and directories | Keep one fact sheet and update every profile from it |
| Sales process | First calls ask what the buyer read; the CRM logs AI claims | Add the question and the field, review monthly |
Then measure monthly with a fixed set of buyer prompts, as in how to tell if ChatGPT recommends your company, and check Bing Webmaster Tools for cited pages. Google counts AI Overviews and AI Mode traffic inside the normal Web report in Search Console.
Frequently asked questions
Do B2B buyers really use ChatGPT to choose vendors?
Yes, at scale. In G2's March 2026 survey of 1,076 B2B software buyers, 51% start software research with an AI chatbot more often than with Google, and chatbots were the top influence on vendor shortlists. In Gartner's survey of 645 B2B buyers, 45% said they used generative AI, mainly to gather information on vendors and products.
How do I get my company on AI shortlists?
Let the assistants' search crawlers reach your site, publish pages that answer buying questions with specifics, including pricing and comparisons, collect recent reviews where your buyers look, and keep your company facts identical everywhere. Then test a fixed set of buyer prompts every month and fix the gaps the answers show.
Should we block GPTBot and other AI crawlers?
Separate search from training. OpenAI says sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, while GPTBot collects content that may be used for model training. Perplexity says PerplexityBot is not used to crawl content for AI foundation models. Google differs: its Google-Extended token covers both model training and grounding in Gemini Apps, so blocking it also keeps your content out of Gemini's grounded answers.
Do we need llms.txt or special schema to appear in AI answers?
Not for Google's AI Overviews and AI Mode. Google says you do not need new machine-readable files, AI text files or markup, and no special schema.org structured data. The OpenAI and Perplexity crawler pages cited here deal with robots.txt access, not extra files. Clear, answer-first pages matter more.
Get on the shortlist, then win the validation call
Getting named by AI assistants is half a marketing job and half a sales job. Our SEO and AI visibility service works on how search engines and AI assistants find and describe you; Full Marketing plans are on our marketing pricing page.
On the sales side, sales strategy development turns the validation-first call into a written playbook. Sales Engine plans start at 1,750 euros a month with five qualified meetings a month guaranteed; see our pricing page.
Sources and further reading: G2, press release, April 2026; G2, The Answer Economy report; Gartner, buyer survey press release, May 2026; TrustRadius, B2B Buying Disconnect Report 2026; Google, AI features and your website; Google, common crawlers and Google-Extended; OpenAI, crawler overview; Perplexity, crawler documentation; Bing Webmaster Blog, AI Performance report; Aggarwal and others, GEO paper, KDD 2024.