SI Signal & Funnel
Growth Experiments

Funnel Analysis: Find the Stage That Changed

Signal & Funnel · Updated 2026-07-19 · 4 min read
Funnel Analysis: Find the Stage That Changed
SummaryRun funnel analysis by defining the eligible population, observable stages, required order, identity unit, exclusions, and maturation window. Show counts and aligned transition rates, distinguish completed, in-progress, ineligible, and alternate paths, and validate events and joins before diagnosing. Segment only for a declared question, protect privacy and small groups, and treat a changed stage as a clue that generates explanations or experiments—not as proof of one cause.
Contents

Define a funnel as eligible stage transitions

A funnel analysis compares how an eligible population moves through a defined sequence of observable stages. Start with the decision and journey. Then specify each stage, entry rule, completion rule, order, identity unit, and time window.

Do not select events merely because they form a declining sequence. A funnel must model a defined process.

Build mutually understandable stages

Use stages that represent meaningful progress: eligible visit, product view, valid form start, accepted submission, qualified review, activated account, or another sequence appropriate to the work. The conversion guide helps distinguish a primary outcome from diagnostic actions.

Document whether stages must occur in order, whether they can repeat, and whether people may skip one. Define exclusions for tests, duplicates, bots, internal traffic, cancellations, and ineligible records.

Align the population and window

Choose whether the funnel follows a starting cohort forward or counts stage activity within one reporting period. These methods answer different questions. A same-period view can compare operational volumes; a cohort view can preserve the people who entered together while later stages mature.

State the unit—person, account, session, order, or lead—and do not switch it mid-funnel. Identity matching carries uncertainty and privacy duties.

Calculate each transition

For a stage transition, divide the number of eligible units reaching the next stage by the number eligible at the prior stage, using aligned rules. Show counts. A high rate on a tiny base can look persuasive while representing very little work.

Avoid calling everyone who did not advance a “drop-off” when some remain in progress, become ineligible, or take a permitted alternate path. Name those states separately.

Validate before diagnosing

Check event firing, order, duplicates, consent states, time zones, delayed imports, and joins. The data-quality checklist helps distinguish a real stage change from a change caused by tracking implementation.

Never bypass consent or collect unnecessary identity to complete the funnel. Follow current privacy, security, contract, platform, and sector rules under the relevant official regulator guidance named below and qualified local counsel.

Segment one question at a time

Compare relevant segments such as channel, market, device, new versus returning, or campaign only when definitions and sample support the question and the use is lawful. Avoid repeated slicing; declare the comparison before reviewing results.

State whether segment membership was known before the outcome and protect small or sensitive groups. Do not infer protected or sensitive traits for marketing analysis without valid purpose and qualified review.

Turn the changed stage into a test

Identify where counts or rates changed, then list plausible explanations: audience mix, offer, message, latency, operations, price, eligibility, tracking, or seasonality. Evidence may rule some out; the funnel alone does not select a cause.

Use cohort analysis when maturation or acquisition period could explain the pattern. Propose the next diagnostic or experiment with an owner and decision.

Report the affected stage, magnitude, limits, explanations, and next check. Use the funnel to narrow the investigation without presenting the narrowest stage as proof of a cause.

Preserve alternate paths

Record meaningful exits, skips, offline completions, and repeated stages rather than forcing every eligible unit through one ideal sequence. Report how each is treated in the denominator. When the product journey changes, version the funnel definition and mark the date so the historical break remains visible.

Official rule sources

Data-protection and direct-marketing duties depend on jurisdiction, data, purpose, and message. Check the current official source relevant to the people and activity: the European Commission data-protection portal for EU scope, the UK Information Commissioner's Office direct-marketing guidance updated 28 April 2026, the California Privacy Protection Agency laws and regulations for California scope, and the U.S. Federal Trade Commission CAN-SPAM guide for U.S. commercial email. These official pages do not determine whether a rule applies to a specific business. Also check current platform documentation and contracts, and use qualified local privacy or legal counsel for consequential decisions.

General marketing education, not legal, privacy, tax, financial, security, or individualized business advice. An independent publication. Not affiliated with any prior owner of this domain.

FAQ

How do I calculate a funnel conversion rate?

Divide the eligible units reaching the next defined stage by the eligible units at the preceding stage, using the same identity and aligned window, then multiply by 100 for a percentage. Show counts too. Define repeats, skipped stages, in-progress records, exclusions, and delayed completion. A rate is not comparable when either stage or the eligible population changed.

Should a funnel use users or sessions?

Choose the unit that matches the journey and decision. Sessions may suit a single-visit path; people or accounts may suit a longer process, subject to lawful and reliable identity. Do not switch units between stages. Document cross-device and shared-account limits, minimize identity collection, and follow current privacy, consent, security, and platform requirements.

Does the biggest funnel drop show what to fix?

No. It shows where fewer eligible units reach the next stage under your definitions. The difference may reflect intent, eligibility, latency, operations, price, audience mix, tracking, or a deliberate alternate path. Validate data, compare aligned cohorts, and investigate plausible causes. Choose a change only after evidence connects the stage pattern to a controllable explanation.

Which official privacy and marketing sources should I check?

Use the official source that matches the people, jurisdiction, data, and activity: the European Commission data-protection portal for EU scope, the UK Information Commissioner's Office direct-marketing guidance for UK scope, California Privacy Protection Agency laws and regulations for California scope, and the U.S. Federal Trade Commission CAN-SPAM guide for U.S. commercial email. Then check current platform documentation and contracts. Qualified local counsel should review consequential decisions.