Projects
Three projects from Backflip, with the numbers and the tradeoffs. Open one for how it actually went.
Backflip · Acquisition funnel → loan application
Running the funnel on experiments, not opinions
ProblemFunnel decisions ran on conviction, and the biggest one, how much rate information to show applicants, had two senior stakeholders on opposite sides.
ImpactRan 10+ A/B tests against measured drop-off, then settled the rate question with a metric fixed before launch that neither side could reinterpret.
- Conversion 1.5% → 2.4%, on 10% less spend
- 10+ A/B tests in one quarter
Backflip · Retention & vendor cost
Cutting the right thing
ProblemAnalyzer, the deal-analysis tool investors relied on, carried a large data bill and no revenue line. In a cost review it read as an obvious cut.
ImpactShowed it drove 6.7× retention, traced the spend to an automated inbox-scanning feature instead, cut that, and repriced the contract. Product kept, cost down 90%.
- 85% more applications per user
- Data cost −90%
Backflip · Budgets, draws & disbursements
Rebuilding the system we were renting
ProblemEvery construction budget and draw lived on a third-party platform we paid heavily for and could not improve.
ImpactRebuilt budgets, draws and disbursement in-house, retired the vendor, and cut payout from 3.5 to 2.1 business days.
- Vendor retired, ~$275K/yr
- Draw payout 3.5 → 2.1 days
- Budgets in-platform