Building a CRO Experimentation Program From Zero
A growth-stage marketplace with no experimentation capability, where the binding constraint was supply rather than demand.
- Product
- Managed marketplace for refurbished e-bikes
- Model
- Buys, refurbishes, and resells e-bikes across three European markets
- Timeframe
- 3 months
Results
What we did
- Framed a conversion brief as a supply problem and descoped the demand storefront
- Implemented A/B experimentation tooling (GrowthBook) alongside a component library, so tests could be built and shipped without bespoke engineering each time
- Wrote the operating model the team runs
- Experiment build guide and results guide
- Pre-sprint ticket audit cadence
- A definition of ready, so nothing enters a sprint half-specified
- Set supply-side guardrails distinct from the demand funnel's, and held that line when seller-flow tests were judged on buyer metrics
- Ran a competitive teardown across car, electronics, and bike resale marketplaces to source the hypothesis backlog
- Reconciled two conflicting offer-acceptance rates into one number for the CEO
Most teams that want a testing program buy a tool and stall, because the tool was never the constraint. The constraint is a backlog worth testing, a build path that does not need bespoke engineering per test, and an agreed bar for what counts as a win. Those three got built alongside the tooling, which is why the cadence held at four experiments every two weeks rather than spiking and dying. Early results are landing now and the compounding read comes at the end of the program.
We help consumer apps and products reach their next phase of growth.
More case studies
A Full-Funnel Audit to Hit Target LTV:CAC
A CRO Program That Consistently Beats Benchmarks
Overhauling the Acquisition Engine for a Profitable LTV:CAC
Scaling Meta Ads While Increasing Spend Efficiency
Growth notes, in your inbox
New writing on subscription growth and product-led strategy when I have something worth saying. No spam.