Proving demand before the build.

Demand is proved by what people do, not what they say.

Bupa wanted to launch a preventative-health proposition built on genomics — polygenic risk scoring, medication checks, condition screening. Real build cost, real reputational exposure.

The category is littered with expensive failures: scientifically impressive tests that sold on curiosity, spiked, then stalled. One genomics proposition, behind a sophisticated and well-partnered distribution strategy, still sold fewer than a thousand consumer tests — behavioural proof that genomics without a clinical trigger doesn't convert.

The core risk was the oldest one in product — building something no one buys. The brief was to retire that risk before the build, with behavioural evidence benchmarked against the market and against itself.

Motivation does not equal action. Even proactive people who said they'd act often didn't.

Interviews and surveys shaped the hypotheses. Clicks, dwell and comprehension tested them — each feature run as its own creative, so demand attributed to the feature rather than to the Bupa brand.

• 01Problem research. 24 in-depth interviews (16 B2C, 8 B2B) and a survey of 100–120 screened target-audience members, surfacing the objections any proposition would have to survive: price, privacy, actionability and the fear of an unchangeable diagnosis.
• 02Market landscape. We scanned the global market to inherit the patterns that separate traction from failure before spending anything. Categories anchored to a live problem built durable businesses; players selling the science showed the failure signature — no meaningful funding, micro-teams, minimal revenue.
• 03Proposition design. Structured ideation generated six raw propositions, then narrowed on evidence to four costed value propositions — each wrapping PRS or MedCheck inside a recognised condition and a clear next step, and each defined precisely enough to test.
• 04Search-trend benchmarking. UK monthly search volume across around 50 query variants covering the four conditions, benchmarking latent demand condition by condition before a single ad ran.
• 05Smoke test. A two-stage demand funnel: four paid social carousels into a live, instrumented landing page. We measured impression to click-through (resonance), website attention (comprehension) and onward clicks (intent), across three rounds and 110,000+ tested impressions.

Leading with the condition beat leading with the technology, and action beat education. The message we prioritised on earned around three times the click-through of the creative that led on the science behind it.

The comprehension layer then caught a false positive. For one condition the clicks looked healthy, but sustained attention ran roughly a hundred times below the rest — and the drop-off showed where in the journey people lost the thread. That told Bupa which part of the proposition to fix rather than which one to abandon. A click-only test would have shipped it.

Bupa enters build knowing which feature to lead with, for which audience, with which message — each backed by behaviour, benchmarked against the market and against rival messages.