Don’t blend discovery calls with customer research

Over eight weeks this summer, I ran a product-market fit study for a UK B2B SaaS business: 14 scripted research interviews and 25 discovery and account calls, 39 conversations in all. I reported the two sets separately from start to finish, each with its own denominator, and that decision exposed the finding that redirected the go-to-market plan.

One pain point, two answers

An administrative pain point came up in 1 of 14 interviews (7%) and in 10 of 25 discovery calls (40%). Blended, it reads as 11 of 39 (28%), a moderate need across the market. Neither segment looked like that.

The interviews had reached one segment of the ICP and the calls another. For the first group the problem barely registered; for the second it was a buying trigger. A blended figure would have sent product and marketing after an average customer who doesn’t exist.

Two kinds of evidence, two biases

A research interview follows a script with no pitch, against a sample chosen to cover the ICP. A discovery call follows the seller’s agenda, with whichever prospect the pipeline delivered. Pool them and you lose sight of which bias produced the number.

Reporting rules

I gave the board every finding as two counts, never as one percentage.

I also counted prompted and unprompted mentions apart. One theme appeared in 11 of 14 interviews (79%), and in 6 of those it came up only after the interviewer raised it. Reported as 79%, it would have topped the priority list.

The report opened with its own gaps: six interviews had no write-up, and most conversations survived only as call summaries.

The decision it led to

Across all 39 conversations, nobody cited a missing feature as the reason for not buying. Thirteen (33%) cited price, budget or procurement. That held in both sets, so the CEO moved effort from the roadmap to pricing and sales execution.

Before your next round of customer conversations

Tag each conversation as research or pipeline before it happens and keep two tallies. Where the two numbers diverge, segment the ICP before you change the roadmap.

Related: how a 23-interview commercial audit replaced opinion with evidence, or read the case studies.