Funnel Experiments
Part of Improving an underperforming lead funnel
Choosing the next lead funnel fix from evidence
Compare possible funnel fixes against the observed loss, evidence quality and outcome that would justify keeping a change.
Choose the next lead funnel fix by linking an observed loss to a change that could plausibly remove it. Repair a confirmed failure that prevents requests arriving or receiving a reply. For less certain problems, record the hypothesis, the outcome to watch and the evidence that would change the decision.
Put candidates on one decision sheet
For each possible fix, record the affected stage, evidence, proposed change, decision-relevant outcome and uncertainty. This keeps a prominent dashboard number from becoming the plan by default.
| Candidate | Evidence needed | Outcome to review |
|---|---|---|
| Repair a form handoff | Successful submissions missing from the receiving queue | More completed requests reach an owner |
| Clarify a service promise | Repeated genuine requests for work outside scope from that route | More relevant requests under the same assessment rule |
| Change the first response process | Suitable requests with a verified wait beyond the stated promise | More requests receive a relevant reply within that promise |
| Change an offer | Requests and buyer questions suggest the offer misses a real need | More suitable enquiries, not merely more clicks |
These are candidates, not a ranking. A confirmed lost-request route may need immediate repair. An offer change requires evidence that the proposed promise better serves the buyer's need.
Decide what the evidence can support
First check that the gap is real under a stable definition. Review the original requests, exclusions, pending assessments and missing records.
A rate drawn from a few reviewed cases is a weak basis for a broad campaign change. A tracking break may justify repairing measurement while leaving the campaign comparison unresolved.
Next ask whether the proposed fix acts on the observed cause. A new button cannot answer requests left in an unowned queue. A qualifying field may sort requests for an unavailable service after they arrive, while the arrival promise remains misleading. Check the business's own cases before applying either inference.
Consider the cost, reversibility and time needed to observe an outcome. Name dependencies: for example, settle an acceptance-rule disagreement before comparing later enquiry groups. Choose the smallest change that addresses a confirmed problem, or the next check that could distinguish competing explanations.
Record and revisit the choice
A decision entry should name the affected route, the observed requests, the change, its owner, the rule used to assess later requests and a review date. Keep the prior wording and rule date available.
For example, if requests from one service page repeatedly concern work outside the published scope, clarify that limit near the enquiry action and review subsequent requests under the same acceptance rule. This is an illustration, not a finding about a site or a forecast.
Watch the intended outcome and neighbouring measures. A clearer promise may reduce irrelevant submissions while leaving accepted enquiries steady, which could reduce assessment work. More submissions accompanied by fewer accepted opportunities calls for another request review.
If evidence is incomplete, choose a specific information-gathering action: verify delivery, finish pending assessments or inspect a defined group of requests. Record the operating choice until that check is done.
