Question about creating an Average Time to Complete Flow calculated metric

Hi,

Best practice recommendation

For this use case, do not use visits.

Visits are session-based and will undercount completions when:

• a user completes multiple bookings in one session
• users restart the flow within the same visit

So using visits will inflate your average time per completion.

What about events?

Using events is closer to correct, but only if your “Booked Appointment” event is:

• fired once per successful completion
• not duplicated due to reloads or retries
• consistently implemented across devices

If that’s true, then:

Time To Complete OLS Flow / Booked Appointments (Event) is the correct approach

Why your numbers differ so much

• Visits denominator → fewer completions → higher average time
• Events denominator → more granular completions → lower (and usually more accurate) average time

This gap is expected and usually highlights a definition mismatch rather than a data issue.

Best practice (what most teams do)

Instead of relying on visits or raw events, the ideal setup is:

• A deduplicated “Flow Completion” metric (based on the confirmation event)
• Scoped at the correct level (usually event/person depending on your definition)

• Then calculate:

  Total flow duration ÷ number of completions

This ensures you are measuring true completed journeys, not session artifacts

Bottom line

• Visits = not recommended for flow completion analysis
• Events = acceptable only if clean and deduplicated
• Best practice = dedicated “completion” metric aligned to confirmation step