Skip to content
Forrest Miller

Method

How I turn evidence into product decisions.

  1. Start with a real problem

    Look for work people already do and what it costs them. Trip planning takes about 18 hours; ~25 million US adults formally plan trips, and 37% now use AI.

  2. Size the opportunity

    Measure the audience, current alternatives, acquisition path, and operating constraints before choosing a product shape. Bingo incumbents draw ~1.1M monthly visits; Allrecipes alone draws about 104 million.

  3. Run the smallest useful test

    Build only enough to answer the riskiest open question, and instrument the core action before launch. A 12-day referrer audit narrowed 533 unattributed sessions to 10 recoverable visits, so the proposed pipeline stopped there.

  4. Learn from real behavior

    Measure what changed, group the behavior by cause, rerun the same audit, and keep corrections in the record. RecipeStripper's failure rate moved 24% → 19.3%; its public counter was corrected 4,886 → 1,743.

    Impact measurement

    One view across three products.

    The dashboard compares the same questions—traffic, acquisition, pages, and behavior—while keeping each outcome specific: trips planned, games played, or recipes stripped.

    Thirty-day traffic dashboard with TripSapien expanded into acquisition, pages, and activity; BingWow and RecipeStripper remain visible below.
    TripSapien's 30-day acquisition, page, and activity view, with BingWow and RecipeStripper below. Captured August 13, 2026.Open the live dashboard (opens in a new tab)

  5. Decide what happens next

    Set the decision rule before the test, then invest, change course, stop, or wait. 47 → 0 on cold outreach closed the channel; TripSapien's ranking change waits for a full month of keep, remove, and swap data.