Find the constraint: read your funnel like a diagnostician

You cannot break a constraint you have not located, and you locate it with numbers, not instinct. The diagnosis is mechanical, and it is the most important hour you will spend on growth this quarter.

Lay every stage against its benchmark

Write out your funnel as a column of stage-to-stage conversion rates and put a benchmark next to each one. For B2B SaaS the rough reference points are well established: visitor-to-lead sits around 1.4% for SMB and mid-market and nearer 0.7% at the enterprise end, and lead-to-customer overall lands around 2.7%. Mid-funnel, the MQL-to-SQL handoff runs a median near 15% across all B2B, in a band of roughly 12% to 21%, while B2B SaaS specifically tends to run hotter at 18% to 22% and top performers reach 25% to 35%. Then keep going past the sale, because retention and expansion are funnel stages too, and they are the ones people forget to instrument. If you want each benchmark with its source, the visitor-to-lead and funnel metrics playbooks carry the full tables.

Numbers in hand, run two tests. First, which stage is furthest below its benchmark? Second, which stage, improved by a single point, would move revenue the most? Those two answers often disagree, and when they do, revenue-lift wins the tie. A stage can be badly below benchmark yet sit early enough in the funnel that fixing it barely touches the bottom line, while a stage close to benchmark but high in dollar value is the one worth your cycle.

The constraint can sit anywhere

Resist the reflex to assume it lives at the front. The constraint can be front-of-funnel (genuinely no demand), mid-funnel (qualification criteria too loose or response too slow), or back-of-funnel (you are winning deals and then leaking them through churn). For most early-stage B2B companies the real constraint is demand or activation; as a company matures it tends to migrate toward win-rate and then retention. Diagnose where yours actually is. Do not inherit someone else's answer.

Thin data is not an excuse

The most common objection at this point is 'my data is too sparse to know.' You almost never need perfect data. You need stage-to-stage rates against benchmarks and one honest question about which stage moves revenue most. Directional numbers beat optimising on vibes every time. And if you genuinely cannot see your stages, that is not a reason to skip the diagnosis; instrumenting the funnel so you can read it is itself the first move. Wiring up the measurement is the opening play, not a prerequisite you are allowed to skip, and a clean CRM foundation plus working analytics setup is usually what makes the stages legible in the first place.