Why the customer interview is the most important artifact
Every other data source in a commercial due diligence process can be read by both sides. Industry reports, government data, web scraping, company financials. The advantage a sponsor gets from CDD does not come from those sources. It comes from the conversations the seller cannot replicate.
The honest answer to “why did you choose this vendor over the alternatives,” from a customer who has no incentive to make the seller look good, is the most valuable data point in the process. The job of voice-of-market (VOM) research is to gather enough of those answers, structured carefully enough, that a pattern emerges.
What VOM research includes
At 2nd St Strategy, a full VOM program includes some mix of:
- Current customers. Why they bought, what they considered, what would make them switch, what they pay, how satisfied they are.
- Lost and former customers. Often the most informative calls in the whole program. Why they did not choose the target, or why they left.
- Channel partners. Distributors, resellers, integrators, brokers. Whoever sits between the target and the end customer.
- Competitor employees. Where possible, current or former employees at named competitors. Sourcing has to be careful; the conversations are gold.
- Topical experts. Former executives at adjacent companies, retired industry leaders, regulatory experts, technical experts.
- Quantitative surveys. Where the population supports it, B2B or B2C surveys to size the patterns from the interviews.
The four mistakes we see most often
1. Overweighting expert calls because they are easier to schedule
Expert networks make it easy to book 15 retired industry executives in a week. Customers are harder to reach. The temptation is to backfill the call count with experts and underweight customers. The result is a report that describes how the category works (which the sponsor probably already knew) without answering why customers actually buy this particular product.
2. Asking leading questions to confirm the thesis
“You like the product, right?” is the worst possible CDD question, and a surprising number of interview guides ask the equivalent. We open every call with neutral, open-ended questions and only get to thesis-specific probes after the customer has framed the category in their own words.
3. Confusing satisfaction with stickiness
A 9 out of 10 NPS does not mean a customer will not switch. A 6 does not mean they will. The right question is “what would make you switch, and how hard would it be.” That tells you about the moat. Satisfaction tells you about the relationship today.
4. No formal pattern coding
Calls without structured coding turn into anecdotes. The deck ends up with three customer quotes that support the thesis and three that contradict it, and the reader cannot tell which way the weight of evidence points. We code every call on a structured grid so the synthesis is reproducible and the patterns are visible.
How AI changes (and does not change) the work
AI is genuinely useful for three things in VOM research.
- Transcription. Faster, cleaner, with speaker labels. We use a hardened pipeline for this.
- Theme extraction. First-pass coding of dozens of transcripts against a structured framework. A human still reviews, but the time-to-first-draft is meaningfully shorter.
- Pattern matching across engagements. The accumulated transcript corpus across 200+ engagements is itself a competitive asset. We can answer “has any customer in any deal we have done before said anything like this” in a way a one-off CDD shop cannot.
AI does not source the calls. AI does not run the call. AI does not decide whether a customer is hedging. The actual conversation is still the work, and the conversation still requires a senior person who knows what to listen for.
What 2nd St delivers
On a typical engagement, the VOM artifact is a structured chapter in the CDD deck with a verdict on each sub-thesis, the coded evidence behind each verdict, and selected anonymized quotes. The full call memo set lives behind it for the sponsor to read if they want.
Across 200+ engagements we have run 3,000+ expert and customer interviews and executed 80+ B2B and B2C surveys. Each one tightens the pattern recognition for the next.
Get in touch if you have an active process where the question is “what would the customer actually say.”
