06 July 2025
Primary market research to obtain B2B customer insights is hugely difficult, slow, and expensive. But now you can interview a robot instead. Here is how it works.
In B2B decision-making—where purchase cycles are long, stakeholders are many, and data can be scarce—relying on a single source of market intelligence often leaves critical gaps. Synthetic research bridges those gaps by blending multiple research methods, data streams, and expert perspectives into a cohesive whole.
Rather than treating “primary” and “secondary” research as siloed exercises, synthetic research weaves them together, triangulating findings until a clear, actionable narrative emerges.
Interviewing B2B decision makers is nigh on impossible
C-suite executives, procurement people, engineers, and operation people don't have time to participate in the type of lengthy interview needed to illicit their opinions and give-up valuable insights into the buying journeys, decision making criteria, relative brand strengths between competing offers, and opinions.
Even if you have breached the fortress and got through the gatekeepers, their first objection is "I don't have time to waste on this." If you catch them at the wrong moment on a busy day, they can be down right rude. If they do participate they may provide monosyllabic answers. Further, some fear they might be giving away corporate secrets or providing data to a competitor. Interviews rarely cover off all the discussion points.
But AI may provide a solution...
A new twist on synthetic research - large language models to “become” AI-generated customer personas
In recent months, a new twist on synthetic research has emerged: instead of recruiting real customers for qualitative interviews, researchers prompt large language models to “become” AI-generated customer personas and then conduct interviews with these digital twins.
This technique builds on traditional synthetic-persona methods—where you craft profile backgrounds, goals, and pain points from past data—by leveraging advanced LLMs to simulate rich, context-aware responses as if they were real users.
Early adopters describe a two-stage workflow. First, teams ingest whatever real user data they have—survey results, past interview transcripts, usage logs—and fine-tune or prompt-engineer an LLM so it embodies the language patterns, concerns, and decision criteria of a target segment. Next, instead of scheduling human interviews, they run their discussion guide straight against the AI persona. The model “responds” with detailed motivations, objections, and even unprompted anecdotes, giving researchers rapid, scalable insights into how a typical buyer might react to new features or messaging.
Proponents argue this approach offers major advantages:
-
Speed & Scale: You can run dozens of “interviews” in minutes without recruitment headaches.
-
Cost Efficiency: No incentives, transcriptions, or moderator fees.
-
Privacy & Access: You avoid confidentiality issues when exploring regulated or hard-to-reach segments.
FLY MEETS OINTMENT
However, rigorous evaluations—such as the recent ArXiv study “LLM Generated Persona is a Promise with a Catch —warn of systematic biases. If the synthetic persona is trained on unbalanced or outdated data, its “opinions” may amplify stereotypes or diverge from real-world behaviors. As with any synthetic method, the key is to treat AI-driven interviews not as a wholesale replacement for human research, but as a complement: a rapid way to refine hypotheses, sharpen discussion guides, and explore edge-case scenarios before—or alongside—validating with live customers.
Less radical synthetic research - improving on traditional field work
At its core, synthetic research is a methodological mosaic:
-
Secondary sources such as industry reports, analyst papers, published benchmarks, and public data sets provide broad context—market sizing, competitive landscape, technology trends.
-
Primary inputs—surveys, in-depth interviews, focus groups or ethnographic visits—deliver fresh, first-hand insights into specific customer needs, pain points, and decision criteria.
-
Expert validation (internal SMEs, channel partners, academic specialists) tests emerging hypotheses and fills in domain nuances that raw data can’t capture.
In a synthetic study, you don’t simply tack these various findings side by side: you actively reconcile them. Where published data diverges from customer testimonials, you probe further. Where expert opinion conflicts with panel metrics, you dig deeper into methodology. The result is a richer, more reliable picture of the market.
Why synthetic market research matters in B2B
Multi-stakeholder complexity: A decision to buy enterprise software or industrial equipment involves technical users, procurement officers, finance teams, and often C-suite sponsors. No single research tool can capture every perspective.
-
Sparse public data: Niche B2B segments often lack up-to-date syndicated reports. Secondary research tells you “what” is happening broadly, but not “why” in your specific vertical.
-
High-stakes investment: B2B purchases carry big price tags and long payback periods. Flawed assumptions can derail entire programs or lead to costly retrofits.
-
Synthetic research helps you validate and stress-test every assumption—so your go-to-market strategy, product roadmap and value propositions rest on rock-solid evidence.
A Step-by-Step Approach
Define the Research Questions
-
What are the unmet needs or purchase drivers in this segment?
-
How do decision criteria differ across stakeholder groups?
-
Which competitive offerings are winning—and why?
Assemble Secondary Intelligence
-
Pull industry reports, financial filings, analyst commentaries, government data.
-
Catalog key metrics: market growth, adoption rates, pricing models.
Design Primary Inquiries
-
Craft surveys targeting end-users and procurement leads.
-
Schedule in-depth interviews or focus groups to explore motivations and barriers.
Engage Subject-Matter Experts
-
Hold workshops with internal sales engineers, channel partners, technical consultants.
-
Test early findings against their frontline experience.
Triangulate & Reconcile
-
Where a report cites 20% annual growth but your interviews suggest flat demand, revisit both sources: Was your sample skewed? Is the public data dated?
-
Synthesize a unified set of insights, noting any remaining uncertainties or data gaps.
Translate into Strategy
-
Build buyer personas that integrate statistical prevalence with behavioral nuance.
-
Craft value propositions tailored to each stakeholder’s pain points.
-
Prioritize go-to-market investments (content, channel enablement, pilot programs) based on which insights are most robust.
Synthetic market research benefits & best practices
-
Holistic Accuracy: Blended methods reduce blind spots and minimize biases inherent in any single approach.
-
Speed & Agility: You can lean on secondary data for immediate context, then layer in just enough primary research to fill key gaps—avoiding lengthy end-to-end fieldwork.
-
Cross-functional Buy-in: Involving sales, product and finance stakeholders in expert workshops builds shared ownership of the insights.
Best Practices:
-
Pre-register your hypotheses: know what you want to confirm or refute.
-
Maintain an evidence matrix: track where each insight originated and its confidence level.
-
Be transparent about limitations: note where questions remain or the data is inconclusive.
A example of synthetic market research in action
Imagine a SaaS vendor entering the logistics-management space. Secondary reports forecast 15% CAGR in fleet-optimization tools, but say little about actual buyer priorities. Through synthetic research, the vendor might discover:
-
Procurement teams prioritize TCO and integration ease above all.
-
Operations managers value real-time mobile alerts and offline capabilities.
-
SMEs warn that small carriers won’t pay for modules they can’t immediately deploy.
Armed with these layered insights, the vendor refines its roadmap (lightweight mobile app, seamless API), creates tailored collateral (TCO calculator for finance teams), and sets up a pilot-to-paid conversion path for small carriers—accelerating growth while avoiding costly feature bets.
