Custom event conversion rate

Custom event conversion rate is the percentage of users or sessions that complete a specific event you have defined, out of those who reached the step immediately before it. If 1,000 users land on a setup screen and 350 of them fire the event you named workflow_activated, the custom event conversion rate for that step is 35%. The event is something you instrument deliberately, which is what separates it from a stock metric like a checkout completion.

It matters because the moments that decide whether a product succeeds are rarely the ones a tool tracks out of the box. A user reaching activation, connecting a data source, or inviting a teammate often predicts retention far better than a pageview or a sign-up. By defining the event yourself and measuring conversion to it, you get a number tied to the behaviour you believe drives value, rather than a proxy that happens to be easy to collect.

The calculation divides the count of users who fired the target event by the count who reached the qualifying step, then multiplies by 100. The accuracy of the number depends almost entirely on how cleanly you define the event and the step before it.

1. 1

   Target event definition

   The exact event you are measuring conversion to, with clear firing conditions. An event that fires inconsistently, or that means different things in different places, produces a rate you cannot trust. Pin down precisely when it should and should not fire.

2. 2

   Qualifying population

   The set of users or sessions that reached the preceding step. This is the denominator. Choosing all users versus only those who reached the step changes the meaning of the metric entirely, so state it explicitly.

3. 3

   Attribution window

   The time allowed for the target event to follow the qualifying step. A conversion that happens within one session is a different signal from one that happens within seven days. Fix the window so the number stays comparable over time.

4. 4

   Deduplication rule

   Whether you count unique users or raw event fires. A single user firing the event five times should usually count once. Without deduplication, a small group of active users can inflate the rate.

A worked example shows why each input matters. Suppose 2,000 users reach an onboarding step in a week, and within their first session 600 fire the target event. Counted by unique user within the session window, the custom event conversion rate is 600 divided by 2,000, or 30%. If you instead counted all sign-ups for the week as the denominator, including those who never reached the step, the same 600 conversions might read as 12%, telling a very different story about the same step.

A metric tree decomposes custom event conversion rate into the things that determine whether a user completes the event, then traces each one back to the team that can change it. This turns a single drop-off percentage into a diagnosis of where and why people stall.

The first level splits the rate into reach, intent, friction, and trust. Reach asks whether qualified users are even arriving at the step. Intent covers how relevant the step is to why the user came. Friction captures the effort the step demands, from form length to load time. Trust covers the doubts that stop a user proceeding even when they want to. Each branch decomposes further, so a falling rate points to a specific cause rather than a general decline.

KPI Tree attaches RACI ownership to every node, so the team accountable for, say, page load time or copy clarity is named on the branch. When the conversion rate moves, the push goes to the accountable owner of the node that changed, and the verified impact loop checks whether a fix, such as cutting two fields from a form, actually lifted the rate rather than just shifting the drop-off downstream.

Because the event is one you define, there is no universal benchmark. The right comparison is against the step type and against your own trend. A low-friction click converts far higher than a multi-field setup, so judge each event against events of similar effort, not against each other.

Read the rate alongside what happens after the event, not just the conversion itself. A high custom event conversion rate is hollow if the users who convert do not stick around. Pairing it with retention rate or feature adoption rate shows whether you are optimising for a meaningful behaviour or just an easy click.

Improving the rate means reducing the friction between the qualifying step and the event, or raising the intent of the users who reach it. The most common mistake is to optimise the conversion in isolation, pushing more people through a step that does not actually lead to value.

The metric tree approach starts by finding the branch with the largest gap between current and achievable conversion. If the friction branch shows a slow-loading step, fixing performance will lift the rate more than rewriting copy. If reach is the problem, more users arriving at the step matters more than converting the few who do.

KPI Tree connects each branch to the team that owns it. Engineering owns load time and error rate. Product and design own the friction and clarity of the step. Marketing owns the quality of the users arriving. When the custom event conversion rate moves, the accountable owner of the node that changed is notified, so the fix lands with the team that can actually make it.

1. 1

   Leaving the denominator undefined

   Measuring against all users versus only those who reached the preceding step gives completely different rates for the same event. State the qualifying population explicitly or the metric is uninterpretable.

2. 2

   Trusting a sloppy event definition

   An event that fires inconsistently, double-fires, or means different things in different places produces a rate you cannot act on. Validate firing conditions before you read anything into the number.

3. 3

   Ignoring the attribution window

   Counting conversions within a session is a different signal from counting them within a week. Without a fixed window the rate drifts as user behaviour changes, and trends become unreadable.

4. 4

   Optimising the click over the outcome

   Pushing more users through an event that does not lead to retention or value inflates a vanity number. Always pair the conversion rate with what happens after the event fires.