AI buyer persona work: Why real interviews still comes first

If your content is not resonating, the problem is rarely the content itself. The real issue is usually simpler: you do not know your buyer well enough. An AI buyer persona only works when it is built on real interviews first, with AI added on top, not the other way around. 

That gap used to be visible: a weak persona sat in a slide deck, ignored. Today it is easier to hide, since an AI buyer persona tool can generate a polished profile from your CRM and website data in minutes, complete with a name, photo, and list of goals. 

Then your sales team reads it and says that is not who they are actually talking to. This is not an argument against AI in buyer persona work, it is an argument for using it in the right order: AI is brilliant at speed and scale, but it is no substitute for sitting down with a real buyer and asking why. 

Why the AI shortcut is so tempting 

Buyers have already moved. Forrester’s 2026 Buyers’ Journey Survey of nearly 18,000 global business buyers found that 94% used AI somewhere in their most recent purchase, comparing vendors and building internal business cases before a single sales conversation happened. It makes sense to want your persona work to move just as fast, but speed on the buyer’s side does not excuse shortcuts on yours. A persona is only useful if it reflects how your buyer actually thinks, not how a language model predicts they might. 

What AI still cannot tell you 

AI is excellent at processing volume: transcripts, reviews, CRM notes, support tickets, and finding patterns fast. What it cannot do is sit in the room with your buyer and notice what they are not saying. It misses emotional nuance, cannot always tell sarcasm from sincerity, and does not know why a buyer acts the way they do, only that they do. It cannot form the kind of trust that gets someone to say what they really think, rather than what sounds reasonable. 

Prospeo’s 2026 research found that only 18% of companies have actually validated their buyer personas against real data. Most are still built on assumption, whether a human or an AI wrote them down. 

One persona is never the whole picture 

A single, well-built persona still only tells you about one person, not how the purchase actually gets approved. B2B buying happens in a group, often eight to ten stakeholders, checking with each other and looking for reasons to trust the decision before anyone commits. That consensus-building is exactly where trust becomes the real currency in B2B buying, and why our 2026 predictions point to human judgement mattering more, not less, as AI takes over more of the process. Map the buying group first, then build personas for the roles that matter most within it. 

Building an AI buyer persona the right way: research first, AI on top 

The order matters. Start with real, qualitative interviews, map the buying group, and understand goals, worries, and how decisions actually get made. That is the foundation no AI tool can generate for you, because it depends on a real conversation with a real person. Once it exists, AI becomes genuinely useful: turning interview data into queryable tools your team can use, pressure-testing messaging against a persona before a campaign goes live, and speeding up analysis, content creation, and roll-out. 

Our buyer insight and strategy work is built around exactly that sequence: interviews first, activation second. 

How many buyer interviews do you actually need? 

Enough to hear the same pattern repeat, not just once. In practice, that is usually five to eight in-depth interviews per persona, mixing buyers who converted quickly with ones who nearly walked away. Fewer than that and you risk building a persona around one loud opinion. 

How Yuki turned real research into a working content engine 

Yuki, the accounting software platform, shows what this looks like in practice. As Mathieu van Baarle, Sales Enablement Team Lead at Yuki, explained: as a fast-growing, marketing-led business, Yuki needed one shared, accurate view of its buyer, not separate assumptions across sales, customer success, and marketing. 

SPOTONVISION conducted Yuki’s buyer persona research through a series of interviews, building three personas over time as the market shifted from SME accountants to the top 100 accountancy firms. 

What Yuki did next is where AI earned its place. The team used AI video tools to bring the personas to life, having them introduce themselves and run internal training sessions, then combined the persona data with win/loss and competitor analysis to build a content blueprint in Google Gemini, used to check whether ads and content genuinely resonate with the target audience. 

The results speak for themselves: overall conversion up 2%, content creation time cut in half, and sales follow-up emails generated twice as fast. 

Can AI replace the interview itself? 

No. AI can prepare you for an interview, transcribe it, and help analyse it afterwards. It cannot ask the follow-up question that comes from genuinely listening, or notice the pause that tells you a buyer has not finished their thought. As Mathieu put it, persona research matters more as AI takes over more execution, not less. The output is only as good as the data you feed it. 

Wrapping up 

A persona built by AI alone might look complete. It rarely holds up once sales, customer success, and marketing put it to the test. Do the qualitative research first, let it be the real voice of your buyer, then bring in AI to scale what you have learned into content, messaging, and campaigns that actually land. 

Want practical B2B marketing insights in your inbox? Subscribe to the SPOTONVISION newsletter and join thousands of B2B marketers who read it every month.