{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-10-24T22:35:46.881Z"},"content":[{"type":"blog","id":"384f09e3-50c5-4346-b175-933cfc9999f3","slug":"ai-buyer-research-interviews-gtm-playbook","title":"AI Shortlist Interviews: A GTM Playbook for Finding Out What Buyers Asked AI Before They Called You","url":"https://www.koji.so/blog/ai-buyer-research-interviews-gtm-playbook","summary":"A GTM playbook for interviewing recent B2B buyers about how they used AI chatbots to build a vendor shortlist. It covers why this matters now (6sense, G2, Gartner and Forrester data), the five-pass Shortlist Trace method, a nine-question interview guide, setup steps in Koji, metrics and an honest note on limits.","content":"Your buyers are asking an AI chatbot who to shortlist before anyone on your team hears about the deal. This playbook shows a GTM team how to interview recent buyers with AI-moderated conversations, rebuild what they typed, what came back and how they checked it, and fix the part of the journey you cannot see in your CRM. By the end you will have a 20-buyer study, a nine-question interview guide and a decision rule for what to change first.\n\n## Why buyer AI research matters now\n\nYour CRM records the day a buyer first filled in a form. It does not record the weeks before it, and those weeks now start in a chat window. Four recent studies show how much of the decision happens there.\n\n![A person at a laptop in an office researching software options, the way a buyer starts a vendor search](https://images.unsplash.com/photo-1758876017801-f5a892ee460a?crop=entropy&cs=tinysrgb&fit=max&fm=jpg&ixid=M3w5NDUzMDN8MHwxfHNlYXJjaHw5fHxwZXJzb24lMjBsYXB0b3AlMjBvZmZpY2UlMjBkZXNrJTIwcmVzZWFyY2h8ZW58MXwwfHx8MTc5MjgyNDYxMHww&ixlib=rb-4.1.0&q=80&w=1080)\n\n_Photo by [Vitaly Gariev](https://unsplash.com/@silverkblack?utm_source=koji&utm_medium=referral) on [Unsplash](https://unsplash.com/?utm_source=koji&utm_medium=referral)_\n\n**Buyers shortlist before you are in the room.** In 6sense's 2025 B2B Buyer Experience Report, drawn from nearly 4,000 buyer responses, the winning vendor was already on the buyer's Day One shortlist 95% of the time. If you are not on that list, your reps are competing for a deal that is mostly settled. Source: [6sense, 2025 B2B Buyer Experience Report](https://6sense.com/b2b-buyer-experience-report/).\n\n**The shortlist now starts with a chatbot.** G2 surveyed 1,076 B2B software buyers in March 2026. 51% said they begin software research with an AI chatbot more often than with Google, up from 29% in G2's 2025 survey, and 69% said a chatbot led them to choose a different vendor than they expected. Source: [G2 via Demand Gen Report, 2026](https://demandgenreport.com/industry-news/news-brief/half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-g2/52737).\n\n**Buyers check the answer with people, including your reps.** Gartner surveyed 645 B2B buyers in August and September 2025 and reported in May 2026 that 69% prefer to validate AI-generated insights with sales reps. 45% had used generative AI in a recent purchase. Buyers were split on trust: 51% said they were more likely to meet misleading information from generative AI, and 49% said the same of sales reps. Source: [Gartner, May 2026](https://www.businesswire.com/news/home/20260520585188/en/Gartner-Survey-Finds-69-of-B2B-Buyers-Turn-to-Sales-Reps-to-Validate-AI-Generated-Insights).\n\n**The group that checks is large.** Forrester's 2026 State of Business Buying puts the typical buying decision at 13 internal stakeholders and nine external influencers, and reports that generative AI searches are the starting point while buyers lean on colleagues and outside experts to justify the choice. Coverage of the same report says buying groups roughly double in size when the purchase includes generative AI features. Sources: [Forrester, January 2026](https://www.forrester.com/press-newsroom/forrester-2026-the-state-of-business-buying/) and [Digital Commerce 360](https://www.digitalcommerce360.com/2026/01/22/forrester-b2b-buying-ai-2026/).\n\nNone of these studies tells you what buyers in your category typed, which vendors the answer named, or what removed you from the list. That is the gap this interview fills. We compiled the map below from the four sources above:\n\n| What the data says | What it leaves unknown | The interview question that fills it |\n|---|---|---|\n| The winner is on the Day One list 95% of the time | How the list was built | \"Walk me through how you got to your first shortlist.\" |\n| Half of buyers start with a chatbot | The prompt and the answer | \"What did you type first?\" |\n| Most buyers check AI with a rep or a peer | Who they asked and what changed their mind | \"How did you check it?\" |\n| Large groups join and re-check | What each person asked | \"When did others join, and what did each want to know?\" |\n\nThe cost of not asking: you keep spending on content for a funnel that begins somewhere else, and your reps keep meeting buyers who arrive with a half-built opinion that nobody on your team has read.\n\n## What good looks like\n\nTeams that do this well treat it as a reconstruction, not a satisfaction survey. We call the method the **Shortlist Trace**: five passes through the same buyer's story.\n\n1. **Trigger.** What happened the week they started looking.\n2. **First prompt.** The words they typed, as close to verbatim as they can manage.\n3. **The answer.** Which vendors came back, which they expected and did not see, and what was said about each.\n4. **The check.** Who or what they used to verify it: a rep, a peer, a review site, a second chatbot, nothing.\n5. **The cut.** How the list became the Day One shortlist, and who was removed and why.\n\n![Framework showing the five passes of the Shortlist Trace interview: trigger, first prompt, the answer, the check and the cut.](/visuals/buyer-ai-shortlist-trace-framework/buyer-ai-shortlist-trace-framework.png \"Run these five passes in order with every buyer\")\n\nRules from teams that have run this many times:\n\n- **Interview 20 buyers from the last six months.** Take about 10 wins, 6 losses to a competitor and 4 no-decision deals. Fewer than 12 and you are reading anecdotes. Past 25 you mostly hear repeats unless you split by segment.\n- **Recent beats representative.** AI answers change monthly. A buyer from nine months ago describes a tool that no longer behaves the same way.\n- **Ask for the prompt itself.** People summarise (\"I asked about vendors\") and the summary hides the useful part. Ask them to read it out, or to open the history and paste it.\n- **Interview more than the champion.** The finance lead, the IT reviewer and the person who inherited the project ask different questions of an AI than the champion does. Plan on two people per account where you can.\n- **Keep sales out of the invite.** Buyers tell a neutral interviewer what they would soften for a rep they may meet again.\n\n**A decision rule for what to change first.** Once 20 interviews are in, count the buyers who named an AI chatbot as their first stop. This is our working rule, not a published benchmark. If more than a third did, fix how AI describes you before you commission new content: audit your category prompts, correct the facts, and put the claims AI repeats on pages you control. If fewer did, the chatbot is a side channel and the cut (pass five) is where your effort belongs.\n\nCommon mistakes:\n\n- Leading with \"did you use AI?\". It invites a yes-or-no and a rehearsed answer. Start with the trigger.\n- Treating the chatbot's answer as the buyer's opinion. What it said and what they believed are two things. Ask for both.\n- Skipping the losses. The vendors that beat you are the ones the AI named.\n- Running it once. Re-run the same study each half-year, because the answer you are tracking moves.\n\n**Timeline back from your next planning or messaging review (say it is five weeks out).** Week 1: pull the 20 names from the CRM and write the brief. Week 2: send invites. Weeks 3 and 4: interviews complete, and you read the first reports. Week 5: the readout and the first two changes.\n\n## How to set it up in Koji, step by step\n\nYou can try this on the Free plan with your signup credits, using one shared link and text interviews. Voice interviews and private links imported from a CSV need the Interviews plan. A text interview uses 1 credit and a voice interview 3.\n\n1. **Start from the template.** In the dashboard's template gallery, filter for GTM teams and open Buyer Journey Mapping. It already walks the buyer from trigger to signature, so you extend it rather than start blank.\n2. **Rewrite the brief.** Change the problem statement to: \"Buyers in our category build their shortlist before they contact us, and we cannot see how.\" Describe the participant as someone who completed a software purchase in our category in the last six months.\n3. **Add the Shortlist Trace questions.** Use the guide below. Mix open questions with a single choice, a multiple choice and a scale so the report can chart the sources and the trust level.\n4. **Add guardrails.** Tell the interviewer not to pitch, not to name your product first and not to correct what the AI said. Then take the interview yourself in preview, which is not counted or charged.\n5. **Choose voice, text or both.** Voice suits senior buyers who will not type. Text suits people who want to open their chat history and paste the prompt. Let participants choose.\n6. **Invite the buyers.** For a named list, use Import CSV on the Self-recruit tab so each person gets a private link and you can see who has not started. See the [CSV import guide](/docs/importing-participants-csv). For a wider net, share one link.\n7. **Read the report.** Check the source chart first, then the themes on the cut, and open the quotes behind each finding.\n\n![Process showing seven steps to set up buyer AI research interviews in Koji, from the template to reading the report.](/visuals/buyer-ai-research-interview-setup-steps/buyer-ai-research-interview-setup-steps.png \"The seven steps from template to report\")\n\n## Copy this interview guide\n\nPaste this into any tool, or into Koji. The types are the question types Koji supports.\n\n**One-line brief for the \"What do you want to research?\" box:** \"We want to understand how B2B software buyers in our category built their shortlist in the last six months, including which AI tools they used, what they asked, what came back and how they checked it.\"\n\n**Opening line:** \"I'm going to ask how you went about finding and choosing your software. There are no right answers, and I'm not here to sell you anything.\"\n\n1. **(open)** \"Think back to the week this project started. What was going on that made you start looking?\" *Probe:* \"What would have happened if you had done nothing?\"\n2. **(single choice)** \"Where did you start your research? AI chatbot, Google, a colleague or peer, a review site, an analyst report, a vendor you already knew, other.\" *Probe:* \"Why there rather than somewhere else?\"\n3. **(open)** \"If you used an AI tool, what did you type first? Read it out as close to word for word as you can.\" *Probe:* \"What did you change in your second question?\"\n4. **(open)** \"Which vendors did it name, and which did you expect that it left out?\" *Probe:* \"What did it say about each?\"\n5. **(multiple choice)** \"How did you check what it told you? Asked a vendor rep, asked a peer, read reviews, visited vendor sites, a second AI tool, did not check, other.\" *Probe:* \"Which check changed your mind?\"\n6. **(1 to 10 scale)** \"Before you checked, how much did you trust its answer?\" *Probe:* \"What would have moved that number up or down?\"\n7. **(open)** \"Walk me through how you got from that list to your Day One shortlist. Who was removed, and why?\" *Probe:* \"Who made the final cut?\"\n8. **(yes or no)** \"Did you already prefer one vendor before you spoke to any of them?\" *Probe:* \"What made it the favourite?\"\n9. **(open)** \"When did other people join, and what did each of them want to know?\" *Probe:* \"Did anyone ask an AI tool something different from you?\"\n\n**Closing question:** \"What do you wish vendors understood about how you research now?\"\n\n![The Koji interview questions editor with open ended, scale, single choice and multiple choice questions.](/email/shots/shot-question-types.jpg \"The question types you will use for this guide, each with its own follow-up setting\")\n\n## What you get\n\nThe report gives you themes with the share of participants behind each one, charts for the choice and scale questions, and the quotes under every finding. Here is how a GTM team uses it the next day.\n\nIn the **pipeline review**, you can show the source chart: how many buyers started with a chatbot, and how many of them never saw you. In **product marketing and content**, the verbatim prompts become your prompt list for the next AI answer audit (see [AI search visibility measurement](/blog/ai-search-visibility-measurement-2026)). In **sales enablement**, the check question shows what buyers ask reps to confirm, which is the first thing a rep should be ready to answer. In **competitive work**, the cut shows which names fell off and the reason given.\n\n![The Koji report summary with a highlighted claim and a side panel of supporting interview quotes.](/email/shots/shot-evidence-panel.jpg \"Every finding opens to the interview quotes behind it\")\n\n## The same playbook elsewhere\n\nThe method carries across roles. Marketing runs a close cousin, [self-reported attribution interviews](/blog/self-reported-attribution-interviews-marketing-playbook), asking buyers how they heard about you, and [buying group mapping](/blog/buying-group-mapping-interviews-marketing-playbook) to find who else was in the room. Your own team runs the [no-decision loss interviews](/blog/no-decision-loss-interviews-gtm-playbook) on the deals that went quiet. Across all of them the rules hold: ask about what happened, use a neutral voice, and interview soon after the event.\n\n## Why Koji for buyer AI research\n\n| | CRM source field | Post-purchase survey | Rep-run debrief | Koji |\n|---|---|---|---|---|\n| What you learn | Where the form was filled | One tick-box on source | What the buyer tells the rep | The prompt, the answer and the check, in the buyer's words |\n| Depth | One field | Shallow | Depends on the rep | Follow-up questions on each answer |\n| Candour | Not applicable | Thin | Softened for the seller | Neutral interviewer, voice or text |\n| Coverage | Every deal, no why | Low reply rates | A handful | Every buyer you invite |\n| Analysis effort | None | Charts, no reasons | Manual notes | Themes with supporting quotes |\n\n![Table comparing a CRM source field, a post-purchase survey, a rep-run debrief and Koji for researching how buyers use AI to shortlist vendors.](/visuals/buyer-ai-research-methods-compared/buyer-ai-research-methods-compared.png \"How four ways of finding out what buyers asked AI compare\")\n\nFive reasons, each tied to something Koji does today:\n\n- **Follow-up questions on every answer.** Each question can probe, so \"I asked ChatGPT\" becomes the exact prompt and what changed in the second one.\n- **Structured and open questions in one interview.** The source and trust questions are charted, and the open answers are themed.\n- **Voice or text.** Participants choose, and text lets them paste a prompt from their history.\n- **Private links from a CSV.** Interviews and Enterprise plans can send each buyer their own link and show who has not started.\n- **Evidence behind each finding.** Findings show how many participants support them, and the quotes open in a side panel.\n\nFor tools that analyse your own sales calls instead of asking buyers, see [Koji vs Gong](/blog/koji-vs-gong-2026). For what review sites can and cannot tell you, see [G2, Capterra and TrustRadius compared](/blog/software-review-sites-b2b-buyers-2026).\n\n## Questions you might have about AI interviews\n\n**\"Will a buyer talk to an AI about how they used an AI?\"** Some will, and the topic helps: it is a neutral, low-stakes subject compared with a lost deal. Participants choose voice or text, and the invite should say who sent it and how long it takes. We cannot promise a response rate, so send a small batch first and see yours.\n\n**\"Will it feel like another sales call?\"** Not if you set the guardrails. You can tell the interviewer not to pitch, not to name you first and not to defend you, and you can take the interview in preview before anyone else sees it.\n\n**\"What happens to what buyers tell us?\"** Read the [GDPR guide for AI user research](/docs/gdpr-compliant-ai-user-research) and decide how long you keep transcripts before you invite anyone. Voice setup is covered in the [voice interview guide](/docs/setting-up-voice-interviews).\n\n## When Koji may not be the right fit\n\nKoji tells you what buyers say they asked and what they recall coming back. It does not show you what an AI says about you today, and memories of a chat can be wrong. Run a prompt audit alongside the interviews, and use the interviews to choose the prompts. If you sell a handful of seven-figure deals a year to the same few buyers, a few senior calls by a person you trust will beat any interviewer, and relationships matter more than coverage.\n\n## Metrics to track\n\n- Share of recent buyers who started with an AI chatbot.\n- Share of those who saw you in the first answer.\n- Share of buyers who checked the answer with a rep or a peer.\n- Trust in the first answer, on the 1 to 10 scale.\n- Day One shortlist rate: deals where you were on the list at first contact.\n- Response rate by segment, and days from close to interview.\n\n## FAQ\n\n**How many buyer interviews do I need?** About 20 from the last six months, split across wins, competitor losses and no-decision deals.\n\n**Should I interview wins as well as losses?** Yes. Wins show what the AI said that helped you, and you cannot learn that from the losses.\n\n**Do I need the Interviews plan?** Not to start. One shared link with text interviews works on the Free plan. Voice and private links from a CSV need Interviews or Enterprise.\n\n**How is this different from win/loss analysis?** Win/loss asks why you won or lost an evaluation. This starts earlier, with how the list of vendors was built. See our [win/loss interview questions](/blog/win-loss-interview-questions-2026) for the later stage.\n\n**Can buyers share their actual chat history?** Some will paste it, which is why text is useful. Never require it, and say how you will store it.\n\n**How often should I repeat it?** Every six months, with the same guide, because the answers buyers get change.\n\n## Start your first study\n\nOpen the Buyer Journey Mapping template in Koji, add the nine questions above, and send it to your last 20 buyers. For more on how GTM teams use Koji, see the [GTM teams page](/for/gtm-teams). The full set of playbooks for this audience is at [/use-cases/gtm-teams](/use-cases/gtm-teams). For the method behind the later stage of the journey, read our guide to [win/loss analysis](/docs/win-loss-analysis).\n","category":"Use Cases","lastModified":"2026-10-24T06:50:10.73834+00:00","metaTitle":"AI Buyer Research Interviews: A GTM Playbook","metaDescription":"How GTM teams interview buyers to learn what they asked AI before shortlisting. Includes a copyable nine-question guide and a setup walkthrough.","keywords":["buyer ai research interviews","how buyers use ai to research vendors","ai shortlist b2b buyers","b2b buyer journey interviews","generative ai buyer journey","day one shortlist"],"aiSummary":"A GTM playbook for interviewing recent B2B buyers about how they used AI chatbots to build a vendor shortlist. It covers why this matters now (6sense, G2, Gartner and Forrester data), the five-pass Shortlist Trace method, a nine-question interview guide, setup steps in Koji, metrics and an honest note on limits.","aiKeywords":["buyer AI research","AI vendor shortlist","B2B buyer journey interviews","win loss analysis","GTM research","Shortlist Trace"],"aiContentType":"guide","faqItems":[{"answer":"About 20 from the last six months, split across wins, competitor losses and no-decision deals.","question":"How many buyer interviews do I need?"},{"answer":"Yes. Wins show what the AI said that helped you, and you cannot learn that from the losses.","question":"Should I interview wins as well as losses?"},{"answer":"Not to start. One shared link with text interviews works on the Free plan. Voice and private links from a CSV need the Interviews or Enterprise plan.","question":"Do I need the Interviews plan?"},{"answer":"Win/loss asks why you won or lost an evaluation. This starts earlier, with how the list of vendors was built.","question":"How is this different from win/loss analysis?"},{"answer":"Some will paste it, which is why text interviews are useful. Never require it, and say how you will store it.","question":"Can buyers share their actual chat history?"},{"answer":"Every six months with the same guide, because the answers buyers get from AI tools change.","question":"How often should I repeat it?"}],"relatedTopics":["buyer ai research interviews","how buyers use ai to research vendors","b2b buyer journey","GTM research"]}],"pagination":{"total":1,"returned":1,"offset":0}}