Ten-Interview Product Discovery

A wedge from a handful of conversations

Alex Mashrabov describes finding a product wedge from a small number of well-chosen customer interviews, roughly eight to twenty, rather than the thousands the conventional advice implies. In his account, Higgsfield's breakthrough did not come from its low-retention mobile app. It came from eight interviews spanning Hollywood-level directors down to regional commercial producers, all asked the same simple questions about what they had seen, what they had tried, and what their experience was. Eight of the eight named the same missing thing, camera controls. Higgsfield built that in March, added a visual-effects library in April, and that became the initial traction engine.¹

Three method details

Mashrabov surfaces three details that, in his telling, make the approach work.

First, consistency beats volume. When the first eight all converge on the same answer, additional interviews add cost rather than signal. He and the host note that the number founders usually cite is "12 to 20... around 10," not thousands. The operative rule is to interview until the signal stabilizes rather than to hit a fixed count.

Second, interview strangers for unfiltered truth. Higgsfield deliberately talked to people it had no close relationship with, on the reasoning that friends and existing fans give flattering, biased feedback while strangers give the real constraint. Mashrabov frames this as the practical guard against the most common discovery failure, sampling your own fan base and mistaking politeness for demand.

Third, hire the best interviewees. Higgsfield hired four of the eight, converting a one-time discovery exercise into a permanent internal feedback loop and a team he describes as roughly forty percent creators and forty percent engineers. His stated view is that "the best creative-AI products are built in symbiosis between engineers and creators," so bringing the interviewees inside institutionalizes the customer's perspective rather than treating discovery as a phase that ends.

Where it fits and where it strains

Mashrabov presents the pattern as removing a common excuse for not starting: discovery is a few days of conversations, not a months-long survey program. In his telling it is the input that makes a small, fast team viable, feeding directly into the kind of rapid path to first revenue Higgsfield ran. The stranger-interview rule and the practice of hiring interviewees are the two moves he emphasizes as underused, the first because it protects the honesty of the signal and the second because it makes the customer's mental model a durable internal asset instead of a memory that fades.

Mashrabov also names the limits. The claim that eight is enough holds when the signal converges; in a fragmented market where the first eight disagree, the right number is higher, and the rule remains to interview until the signal stabilizes rather than to always stop at eight. And hiring interviewees, while it hard-wires the feedback loop, can narrow perspective over time as the loop becomes an echo chamber of early adopters, so fresh outside discovery still has to continue. This disciplined, evidence-first posture toward a nascent market is the same instinct that lets a founder defend a narrow vertical against a generic tool, since a handful of deep conversations surfaces the specific workflow the generic product misses.