AFAha Funnel AnalyzerFunnel AnalyzerLOCAL CSV / 14-DAY COHORTLOCAL

Pre-aha conversion evidence

Aha Funnel Analyzer

Turn event rows into a bounded cohort funnel, then investigate the break closest to your North Star event first.

Illustrative sample. No Amplitude, Mixpanel, account, revenue, experiment, or live product data is connected.
Load the bundled events or paste your own CSV.
SIGNUP -> ACTIVATED -> TRANSACTION

The completed result will show counts, conversions, loss, and the first pre-aha investigation.

Worked decision

Fix the edge before the aha event, then work backward.

4 / 4

Every signup reaches activation, so this step is not the immediate constraint.

2 / 4

Only half of activated users transact: two users and fifty percentage points are lost.

First move

Investigate activated -> transaction before changing earlier signup or activation behavior.

Which edge gets the first investigation?

Run the sample, then apply the prioritization rule.

Conversion shows where. It does not explain why.

The tool describes supplied events and ordered stage membership. It does not infer causal friction, forecast revenue, verify tracking quality, or replace an experiment.

Cohort funnel arithmetic and evidence boundaries

Read the explanation

The page explains a three stage cohort funnel: signup, activation, then transaction as the aha event. Its worked example says all four signups reach activation and two of those four transact. The bars show four activated users and two transaction users at sixty pixels per user, giving two hundred forty and one hundred twenty pixels. Signup to activation is four divided by four, or one hundred percent. Activation to transaction is two divided by four, or fifty percent. Cumulative signup to transaction is also fifty percent. These numbers come from the explicitly illustrative example in the HTML, not a connected analytics service or independently processed event file. The earlier edge loses zero users, while the edge immediately before the chosen aha event loses two of four activated users. The bars span zero and one hundred sixty pixels at eighty pixels per lost user. The proposed investigation therefore starts at activation to transaction, using the page rule to work backward from the aha event. This identifies where ordered membership drops, not why it drops. Poor measurement, time window choices or genuinely absent behavior can produce the same funnel. The page explicitly rejects inferring causal friction, forecasting revenue, verifying tracking quality or replacing an experiment. A local video should preserve these limits rather than treat the recommendation as proof of a causal intervention. The default date fields show July first through July fourteenth, fourteen calendar dates if the endpoints are inclusive. The HTML alone does not prove how its event engine implements temporal boundaries or repeated and out of order events. The mirrored site directory available here lacks the referenced engine, application and stylesheet files. Offline browser controls can accept pasted text while sample loading, analysis and exports remain unbound. The bars compare fourteen shown dates with zero actually executed engine outputs in this diagnostic. Original and proposed geometry, available static drawings and native input behavior are compared separately; narrated arithmetic is not evidence of a functioning import parser, exhaustive tool behavior or actual public publication.

Super generates helpful tools and automates fact-checking across the internet proactively. If you enjoyed this tool, build your own with Super and share it with a friend.