- The question
- What do software buyers ask AI, and what should a company put on its site?
- The short answer
- In a Wynter survey of 200 software buyers, published in Growth Memo on 9 November 2026, 99 of 158 people who already had vendors in mind said AI added another to their shortlist, and 33 of 40 who started with none said AI gave them one. Of the 126 who reported choosing a vendor, 78 said that vendor was in the first answer, and 75 of those 78 still asked follow-up questions. The follow-up topics reported most often were pricing (57.5%), comparisons (54.5%), integrations (46%) and setup (40%). After the chat, 145 of 200 visited a vendor website. The survey is self-reported, not representative, and the author discloses being an investor in Wynter. Our read: put those four topics in plain text on crawlable pages, because a first mention is not the end of the evaluation.
TL;DR: On 9 November, Growth Memo published an analysis of a Wynter survey of 200 software buyers. Respondents said AI often added a vendor to their shortlist. 75 of the 78 who said their vendor was in the first AI answer still asked follow-up questions. Across the sample, the topics picked most often were pricing, comparisons, integrations and setup. After the chat, 145 of 200 visited a vendor website. The survey is self-reported and not representative, and the author discloses being an investor in Wynter. Our read: put those answers in plain text on pages a crawler can fetch. A mention in the first answer is not the close.
The likely default: treat an appearance in an AI answer as the win, and leave price, integrations and setup for a later call.
The people in this survey did not stop there. They kept asking, and most of them still came to a vendor's site.
What the survey found
The write-up covers 200 responses to a 13-question Wynter survey. Every participant's industry profile included SaaS or software. The percentages are unweighted. Results are self-reported and, in the author's words, "not representative of all B2B buyers." Some personal or non-purchasing accounts were left in. Prompt wording may have been recalled rather than copied. A reported demo or signup is not a confirmed purchase, and the survey cannot show whether a vendor that AI introduced was the one eventually chosen, or that AI caused the choice. The author discloses being an investor in Wynter.
AI reopened the shortlist
Before the chat, 158 of 200 respondents (79%) already had at least one vendor in mind, 40 (20%) had none, and 2 (1%) gave conflicting answers.
Of the 158 who already had candidates, 99 (62.7%) said AI added another vendor to the shortlist. Of the 40 who started with none, 33 (82.5%) said AI gave them a shortlisted vendor. The two groups were not at the same stage: about two-thirds of those without candidates were still exploring the category, compared with about half of those who already had a shortlist.
A brand showing up in an answer, making a shortlist, and getting a demo request are different outcomes. The write-up says it cannot tell whether a vendor AI introduced was later chosen or got a demo or signup.
The first answer was not the end
Of the 126 respondents who reported choosing a vendor, 78 (62%) said that vendor appeared in the first AI answer, and 43 (34%) said it appeared after follow-ups. That leaves 5 of the 126 in neither group.
Early inclusion did not end the conversation. 75 of those 78 still sent follow-up messages.
The write-up says 41.5% of participants sent three to five messages after the first prompt, and that one-third sent six or more. One example quoted in the memo shows a buyer using the chat to cut vendors, not only to collect names: a request to drop any vendor without a specific marketing-automation integration.
What they asked next
The most frequently selected follow-up topics in the sample were:
- pricing, 57.5%
- comparisons, 54.5%
- integrations, 46%
- setup, 40%
The write-up does not state the denominator for those four rates in the text. We are repeating them as reported, not inferring a count.
People who sent more follow-ups also selected more topics. Respondents with one or two follow-ups selected an average of 2.4 topics from the survey's checklist, compared with 5.0 among those with ten or more. That is a description of longer chats, not a claim that a longer chat caused a purchase.
They still visited the website
After the AI chat, 145 of 200 respondents (72.5%) visited vendor websites, and 96 (48.0%) used Google. Other reported next steps included asking peers (31.0%), searching Reddit (29.0%) and checking G2 or other review sites (23.0%). The survey did not record what people searched, how they reached a vendor site, or in what order they did these things. A website visit could itself have been the signup step.
67.5% of the 200 used more than one AI tool for the research. The tools they selected were Claude (71%), ChatGPT (62.5%), Gemini (38.5%) and Google AI Overviews or AI Mode (15%). Those groups overlap. The write-up says the sample is not representative of all B2B buyers.
What it means for a business
This section is our read, not the survey's.
If buyers use AI to reopen a shortlist, then add follow-up tests, the page that matters is the one that can be quoted when they ask about price, a comparison, an integration or setup. If those answers live only in a deck or a call, the chat has nothing solid to repeat, and the buyer who then visits your site still has to hunt for them.
You do not need a public price list to be clear. You need a complete, direct passage that says how you work, what you need from the customer, what you integrate with or run on, and how a first engagement starts. Scatter those facts across a homepage slogan, a footer and a PDF, and you make the model and the visitor do the assembly.
The other number worth sitting with is the site visit. Most people in this sample reported visiting a vendor website after the chat. The survey does not say why they went or what they did there. That is a reason to write the follow-up answers for a human reader as well as for an AI assistant.
It is also a reason not to over-read a single mention. Many of the people who chose a vendor said it was in the first answer. 75 of those 78 still asked follow-up questions.
What to do this week
- Write a plain-text answer for pricing or commercial terms. If you do not publish a price, say how pricing works, what a first conversation covers, or what changes the fee. Put it on a crawlable page, not only behind a form.
- Say what you compare against, in your own words. Who you are for, who you are not for, and what you do and do not include. Buyers asked for comparisons. If you leave a vacuum, the chat will fill it from somewhere else.
- List the tools, platforms or environments you work with. Integrations were a common follow-up, and one quoted prompt cut the list for the lack of a single connection. Name them in ordinary language, near the top of that section.
- Describe setup or onboarding in one complete passage. What happens first, what you need from the customer, and how long the first stage usually takes if you can say so honestly. Do not invent a timeline you do not keep.
- Check that a visitor who arrives after a chat can find the same answers. 145 of 200 people in this survey said they visited a vendor site after the chat. If the plain-text version and the human-readable version disagree, the second visit undoes the first answer.
- Do not treat one engine as the whole market. Two-thirds of this sample used more than one AI tool, and Claude was the most used. If you test how you appear in answers, use more than one assistant and treat the result as a sample, not a ranking.
None of this requires a new format, a special file or a campaign to seed answers. It requires the facts a buyer already asks for, written once, in the open.
Sources
- How software buyers use AI, Growth Memo, 9 November 2026