How attribution actually works

The single most load-bearing claim Olivio makes is "every euro attributed." If that claim is fuzzy, the whole pitch collapses into another marketing-software dashboard. So we're being explicit about how the math works — before the first founding partner signs.

Baselines are frozen on day one. When a gym joins, we take a 30-day snapshot of MRR, churn rate, occupancy, day-pass conversion rate, referral rate. That snapshot doesn't move. Every subsequent measurement compares against it, not against an industry benchmark, not against last week. The counterfactual is the version of the gym that existed the day before we started.

Six categories. Six rules. Each action Olivio takes is tagged at execution time with one of:

• Churn save — credit when the next monthly invoice clears (not when the member books one class).
• Empty-spot fill — credit on paid bookings only. Credits don't count as new revenue.
• Day-pass conversion — first 3 months of the resulting membership, 14-day attribution window.
• Lapsed reactivation — first 3 months of the returned member.
• Referral — first 3 months of the referred member.
• Operational save — one month at 0.5× value.

Holdouts inside each gym. Every campaign holds out ~25% of eligible members as a control group. The causal lift is the intervention return rate minus the control return rate. Without holdouts you can't separate "Olivio worked" from "they would have come back anyway."

The dispute mechanism. Owners can flag any attribution claim. For the first 90 days we eat every dispute — assume the owner is right, refund the attribution. After 90 days we analyze the patterns and refine the rules. The point is to start conservative, not generous.

The whole methodology is documented in a spec we'll publish before any partner signs. If a third-party stats person finds a hole in it, we'd rather hear that now than after the engineering is built around it.