Definition:
In digital analytics, a funnel leak is a stage in a journey or funnel where some of the people who completed one step do not reach the next within the conditions defined for the analysis.
The drop identifies where measured progression stops, but does not by itself prove that an error exists or identify its cause. Some people may leave because of technical or experience problems, while others compare alternatives, postpone a decision or do not meet the process criteria.
Table of contents
How a funnel leak is measured
Measurement requires the steps in the conversion funnel, the events representing each step, the analysed population and the permitted time for progression to be defined.
Between two consecutive steps in a closed funnel, the abandonment rate can be calculated by subtracting the people who reach the second step from those who completed the first, dividing the result by those who completed the first and multiplying it by 100.
For example, if 1,000 people complete one step and 600 reach the next, 400 are lost and the abandonment rate between the two steps is 40%.
The result depends on the funnel configuration. It may change according to whether entry at intermediate steps, repeated events, skipped steps, cross-device journeys or different time windows are permitted. Instrumentation errors can also produce an apparent leak.
Funnel leak scope
A funnel leak is the interpretation of lost progression within a defined journey. The abandonment rate quantifies how many people do not move from one stage to the next.
An exit identifies the last page or screen recorded in a session. It may occur after the objective has been completed successfully and is not necessarily a leak. Bounce is a session metric whose definition depends on the analytics platform and does not by itself identify a funnel problem.
A high rate can therefore locate a stage that merits investigation, but does not prove that the page, message or user is the cause.
Causes that may explain a funnel leak
Causes need to be treated as hypotheses until evidence supports them. Common groups include:
- Measurement: Missing or duplicated events, tagging changes, identity problems or cross-domain and cross-device journeys can distort progression.
- Technical operation: Errors, slow performance, incompatibilities, validation failures or interruptions in payments and forms can prevent the process from continuing.
- Offering and conditions: Price, additional costs, delivery times, requirements or the absence of a suitable option can change a decision.
- Content and usability: Ambiguous instructions, insufficient information, confusing navigation or poor visual hierarchy can make the next step difficult.
- Intent and context: Some people explore, compare, become distracted, change devices or discover that the offering does not meet their needs.
The same drop may combine several causes, and different segments may leave the same stage for different reasons.
Funnel leak management
Analysis begins by checking that events, denominators, identities and time windows are correct. Progression can then be compared by device, browser, channel, market, user type and period to identify patterns.
Technical logs, usability tests, support enquiries, surveys, interaction maps and session recordings provide complementary evidence. None of these sources explains every person’s motivation on its own.
Once a probable cause has been formulated, CRO can be used to prioritise changes and evaluate their effects. An A/B test can compare alternatives when sufficient traffic and a testable hypothesis exist, but errors, revenue, lead quality and other side effects also need to be monitored.
Not every leak needs to be removed. In some processes, excluding unsuitable contacts, preventing incorrect purchases or allowing a person to leave an action they do not wish to complete may be an appropriate outcome.