Together with Matt Wilkinson, we unpack why buyer insight disappears inside companies and how “organizational gravity” quietly turns good strategy into safe messaging that sales will not use. We also show how Persona AI and synthetic customers can keep real customer needs present through approvals while staying grounded in voice of customer and testing.
• Customer presence as the real problem, not missing data • How customer insight gets trapped in VoC folders and CRMs • Organizational gravity pulling messaging back to “safe” internal language • Early signals of drift when sales drops marketing materials • Why generative AI makes interactive personas possible now • Building Persona AI with digital footprints, interviews, and research • Defining synthetic customers and grounding them with context • Validating persona outputs with message testing and real buyers • Two key risks: AI sycophancy and overreliance on AI • Modeling buying roles instead of one “buying committee” persona • Procurement and finance concerns as common deal blockers • Citation compression, brand authority, and machine-readable content • Designing journeys where AI helps but humans stay accessible
How Persona AI Turns Customer Data Into Usable Insight
Most B2B marketing teams do not suffer from a lack of customer data. They suffer from a customer presence problem. Research lives in folders, voice-of-customer notes sit untouched, and CRM fields never turn into insight that shapes real decisions. The result is predictable: the people who approve content often have the least proximity to buyer pain, language, and urgency. As messaging moves through reviews, claims get softened and specificity disappears, until sales stops using the materials because they no longer match live conversations. Fixing this starts by operationalising insight so the buyer’s reality shows up in every draft, every approval, and every campaign decision.
A useful way to explain the drift is “organizational gravity.” Even strong strategy gets pulled back toward safe, internal wording that sounds acceptable but does not land with customers. You can spot the first signal by comparing what the research clearly says with what finally ships, then looking at response rates, sales enablement adoption, and early sales calls. If the commercial team keeps improvising and abandoning marketing assets, alignment is already broken. This is especially common in regulated environments like life sciences marketing, where legal and regulatory review steps can unintentionally optimise for risk avoidance instead of buyer understanding.
Joeri Billast and Matt Wilkinson it the Sintra Synergies retreat
Generative AI changes what is possible because personas no longer need to be static documents. With custom GPTs and deeper research inputs, teams can build Persona AI chatbots that represent key customer roles using grounded context: LinkedIn signals, audience questions, brand sentiment, interview transcripts, and quantitative studies. The goal is not to replace human customer research, but to keep the buyer “in the room” as content moves through complex workflows. Teams can query the persona at each revision to see what a change does to clarity, credibility, and relevance. Done well, this becomes an iterative system that updates as market signals change, improving consistency across account-based marketing, demand generation, and product marketing.
The synthetic customer idea makes this even more practical, but only if you validate it. Treat it like a hypothesis: test persona-driven messaging against real outcomes, then feed results back into the model. Proof points such as higher cold-email engagement and more marketing-qualified leads can show directional value, but the risks are real. Ask leading questions and you get flattering answers, which is classic AI sycophancy. Rely too much on the tool and you stop listening to actual customers. As AI also becomes a buyer-side tool, marketers must make content machine-readable and brand-forward, while still making it easy to reach a human. Trust, authority, and personal branding matter more when the first impression happens inside an AI result, not on your website.
