What should count as a completed checkout?
Start with two events, which are named records of actions. Record Checkout opened when the form appears and Order confirmed only when your app has saved a successful order. A tap on Pay can happen even when payment fails.
Use those two events as the steps in a funnel, a report that shows how many people reach each step. Pick a time window that fits the task. If completion can happen the next day, a one-hour window will leave some successful orders outside the report.
How can you check the funnel before using real orders?
Check two fictional journeys in a separate development project. One finishes checkout and the other gets a deliberate decline from the isolated test payment connection. That gives you a result you can predict before looking at the chart.
- Check the test connections before opening the appConfirm that the browser and server tracking in the disposable app use the development project token and that any forwarding or third-party collection stays inside the test setup. Confirm that checkout reads and saves only fictional orders in disposable storage, uses a test payment connection, and sends no customer receipts. Leave automatic collection and session replay off for this exercise. If you cannot verify the routes, work through both journeys on paper.
- Compare the unfinished and finished journeysUse two fresh browser sessions and fictional accounts. Open checkout once in each. Complete the test payment in one, then use your payment provider’s documented test decline in the other. Open Mixpanel’s Events report in the development project and filter to the fictional accounts. Expect two Checkout opened events, one Order confirmed for the successful account, and zero Order confirmed events for the declined account. Then check that the funnel shows two starts and one finish in the selected time window.
- Decide what the test provesIf the declined test payment creates Order confirmed, fix when the app records the event. A clean test tells you that these two journeys are measured as intended. It cannot tell you why customers abandon checkout.
How does Mixpanel know which actions belong together?
On the web, Mixpanel can start with a browser identifier and link activity when the app supplies a known account ID. Use a stable, opaque ID made for the account. Reset the browser tracker when the person signs out so the next person on that browser starts separately.
Choose which details to send before collecting anything. A broad label such as checkout type can help, while an email address or payment details are unnecessary for this funnel. Session replay is a separate capture setting that records interactions on a page. Leave it off until you can verify that sensitive pages and inputs are excluded.
What counts toward the plan, and what can you keep?
Mixpanel Free lists up to 1 million events per month. A person can create several events in one visit. Estimate how often each tracked action happens and include automatic collection if you enable it. Growth lists its first 1 million monthly events included and charges for higher volume.
Export the underlying event records with the Raw Event Export API. Keep your event definitions and funnel settings alongside them so another tool can rebuild the report. Raw exports can contain duplicates even when reports remove them. For a user deletion request, use the documented privacy workflow and account for any copies you already exported.