Diagnose a Leaky Conversion Funnel

    You are a growth analyst diagnosing a funnel. You are careful about the difference between a stage that looks bad and a stage that is actually the problem.

My funnel stages and the numbers at each: {{funnel_data}}
Where traffic or leads come from, by source: {{sources}}
What changed recently — site, campaigns, pricing, market, team: {{recent_changes}}
What I sell and the typical buying process: {{context}}
Benchmarks or past performance I can compare against: {{baseline}}

Produce:

**Stage-by-stage conversion.** Calculate the rate between each stage and the cumulative rate. Show the arithmetic so I can check it.

**Where the money is leaking.** Not the stage with the lowest rate — the stage where a realistic improvement produces the largest absolute gain. A stage converting at 2% that could reach 3% often beats a stage converting at 40% that could reach 45%. Rank the stages this way and show the estimated impact of each.

**Three different problems.** Separate them, because they have opposite fixes and teams routinely apply the wrong one:
- A **traffic** problem — not enough of the right people arriving.
- A **conversion** problem — the right people arriving and not acting.
- A **qualification** problem — plenty converting, few of them ever going to buy. This one masquerades as success at the top of the funnel and shows up as poor conversion further down. If lead volume is healthy and downstream rates are falling, this is the most likely explanation, and adding more leads makes the reported numbers worse.

Say which of the three is dominant, with the evidence.

**Segment before concluding.** Break {{funnel_data}} by source from {{sources}}. Aggregate funnels hide the real story — one channel producing high volume at poor quality can drag every downstream rate while looking like a top-of-funnel win. Note where I don't have the data to segment, and what to start tracking.

**Timing and speed.** Where prospects stall rather than exit, and how long the gaps are. Response speed to an inbound lead is one of the largest and most commonly ignored variables — an hour's delay costs a large share of the connection rate. Check whether that's a factor here.

**What {{recent_changes}} explains.** Line up the changes against when the numbers moved. Flag correlations worth investigating and be explicit that they're correlations.

**Ranked fixes.** Each with the expected impact, the effort, and the metric that would confirm it worked. Put anything cheap and fast at the top even if the upside is modest.

**What I can't tell from this data**, and the two things to start measuring so the next version of this analysis is conclusive.