B2B funnels leak at four transitions: visitor to lead, lead to MQL, MQL to opportunity, and opportunity to close. The largest single leak is usually MQL to opportunity, and the cause is usually follow-up speed rather than lead quality.
Measure before you diagnose
Every team has an opinion about where their funnel leaks and the opinion is frequently wrong. Marketing believes sales does not follow up; sales believes the leads are poor. Both are arguing from anecdote because nobody has instrumented the stages.
Start by measuring conversion at each transition and quantifying the revenue lost at each one. A stage converting at 20% when comparable benchmarks suggest 35% is losing a calculable amount of money, and that figure is what should drive prioritisation rather than whoever argues most persuasively.
Leak one: visitor to lead
Usual causes are message mismatch between the ad or search result and the landing page, forms requesting far more information than the offer justifies, unclear value above the fold, and mobile experiences that were never properly tested.
Form length is the most reliably over-specified element. Every additional field costs conversion, and the honest question is whether sales genuinely uses each field or whether it was added because someone once wanted it and nobody removed it. A whitepaper download rarely needs more than name, business email and company.
Leak two: lead to MQL
This transition usually leaks because the scoring model is wrong rather than because leads are bad. Most models we audit weight job title heavily and behavioural signals barely, which ranks a senior person who has never visited the site above a mid-level evaluator who has read the pricing page five times.
The consequence is worse than the direct loss. When scoring is wrong, sales learns to distrust the MQL queue and starts working its own lists, at which point marketing’s output stops reaching anyone regardless of quality. Rebuilding scoring on what actually predicts conversion in your historical data is an analysis exercise before it is a configuration one.
Leak three: MQL to opportunity
This is usually the largest leak and the cause is usually speed. Leads sit for days before a rep makes contact, by which point the buyer has spoken to two competitors who responded faster.
Response speed is one of the most consistently measured predictors of B2B conversion, and the advantage decays sharply within the first hour. Most teams do not know they have a speed problem because nobody measures speed-to-lead. Instrumenting it is frequently the single highest-return change available, and it is a routing and alerting fix rather than a people problem.
Leak four: opportunity to close
Deals stall in stages that have no exit criteria and nothing alerting anyone. A deal sitting in ‘Proposal’ for eleven weeks is not being worked, but nothing in the system says so, and it continues to appear in the forecast as though it were live.
Defining exit criteria for every stage and alerting when a deal has not moved within a set period surfaces the problem. In our experience most cycle-time improvement comes from removing invisible dead time rather than from selling faster.
Quantify, then sequence
| Stage | Common cause | Typical fix | Speed of result |
|---|---|---|---|
| Visitor → lead | Message mismatch, long forms | CRO testing, field reduction | 4–8 weeks |
| Lead → MQL | Title-weighted scoring | Rebuild scoring on behavioural data | 6–10 weeks |
| MQL → opportunity | Slow follow-up | Routing, alerting, SLA enforcement | 2–4 weeks |
| Opportunity → close | No stage exit criteria | Define criteria, alert on stalls | One full deal cycle |
Sequence by revenue impact and by speed of result. Speed-to-lead fixes are usually first because the effect size is large and the change is quick. Testing on low-traffic pages comes last because reaching statistical significance takes longest.