Week 5 · Proof
Conversion fell and everyone has a theory
Walk down the metric tree. Don't argue at the top of it.
The brief
Tuesday, 08:15, exec staff. Sofia puts the chart up. Conversion rate for deals created in May and June is well below the first four months of the year.
- Dana: "It's the price increase in May. I said so."
- Lena: "It's the new outbound lead-scoring vendor. The leads are garbage."
- Raj: "Are we sure it isn't the stale-stage problem from week 3?"
- Tomás: "Is this deal count or deal value?"
Four hypotheses, one number. This is what the Metric Dependency Tree is for. A drop at the top has to show up somewhere at the bottom: a segment, a stage, a rep, a source. Root-cause analysis is walking down the tree with a query at each level until one branch explains most of the move.
This week: confirm the drop is real (and not an artifact of open deals), then slice it by region × source, by stage velocity, and by rep. Bring the exec team a cause, a size, and a fix, not a theory.
Lessons
Read
The Proof: Product Performance (Weeks 5–6)
Post-launch is where you spend the most effort: usage metrics, A/B tests, and feedback loops that evolve the SLO model and expand use cases.
Funnel Analysis: Overall vs. Stage-to-Stage, Cohorts, Velocity
Three different funnel questions need three different queries: overall conversion (by cohort), stage-to-stage conversion (from the history log), and time in stage (velocity).
Case Study: Multi-Factory OEE & KPI Monitoring
OEE = Availability × Efficiency × Quality. Legacy/manual OEE computation breaks at multi-factory scale; a medallion pipeline fixes latency and drill-down.