Funnel Analysis 101: How to Find Where Users Drop Off
What funnel analysis is, how to build conversion funnels from event data, and how to read and fix drop-offs.
Every product is a funnel whether you measure it or not. People visit, sign up, try the thing, and a small fraction end up paying. Funnel analysis is simply measuring how many users make it from each step to the next - and it's usually the fastest way to find where your product is leaking.
What a funnel looks like#
Pick the 3-5 steps that define success in your product and count the users who reach each one:
| Step | Users | From previous step |
|---|---|---|
| Visited pricing page | 10,000 | - |
| Signed up | 1,200 | 12% |
| Activated (used a core feature) | 480 | 40% |
| Subscribed | 96 | 20% |
Overall conversion here is about 1%. But that single number is nearly useless - the value is in the step-by-step view.
How to read a funnel#
Find the biggest relative drop. Above, 88% of visitors never sign up, and 60% of signups never activate. The signup-to-activation gap is usually the most fixable: these people already raised their hand, then hit friction. Onboarding improvements, better defaults, and shorter time-to-value all attack this step.
Segment before you conclude. An aggregate funnel hides everything. Break each step down by source, plan, or device and you'll often find one segment is broken - say, mobile users converting at half the rate - rather than the whole funnel being mediocre.
Watch it over time. A funnel measured once is a snapshot. The same funnel week over week is how you know whether the onboarding changes you shipped actually worked.
What you need to build one#
Two things: an event for each step, and a user id on every event so you can count people rather than raw events. If you're already logging product events with a user_id, you can build a funnel today - no new instrumentation required. Our SQL cheat sheet has the funnel step-count query if you want to do it by hand, and we cover retention - what happens to users after the funnel - in Retention Curves 101.
The lazy way#
GraphJSON builds funnels from your event data with a few clicks: pick the steps, see the conversion and drop-off between each, and compare funnels across segments or time periods with our A/B funnel tool. The full walkthrough is in the funnels guide.
Written by JR
Founder and builder of GraphJSON.
