Investigate checkout drop-offs with tasks, data and errors

A gap between started and completed checkouts shows where an investigation can begin. A concrete change also needs the observed workflow and relevant technical events. Start with a small question: where does a particular buying task break down on a smartphone?

Define what is being counted first

Define the time window, device type and funnel steps. Compare sessions counted by the same rules; orders and sessions are different units. Shopify’s documentation also distinguishes open and closed funnels. A comparison only fits when entries and step sequences are treated alike. When changing measurement systems, check definitions before assuming behaviour has changed.

A worked investigation example

A fictional drawing-supplies shop wants to examine this task: select a drawing portfolio, enter a delivery address as a guest, choose shipping and continue to payment. All numbers and observations in the table are fictional; they show how different findings can be brought together.

First check whether the payment records fully represent the transaction being examined. Missing entries may also reflect a gap in matching records. If it becomes clear that the form stops before a payment request is sent, adding another payment method does not fix that form problem.

Observe the task on the actual device

Have the task performed in a suitable test environment. Record the device, browser, keyboard state and last understandable step. Can the person tell which field is wrong? Is previously entered information retained? Can they correct the address and continue with the chosen shipping method?

Payment review complements this observation. Shopify describes errors arising from bank decisions, payment services or shop configuration, for example. A general on-screen message does not establish which case applies. Match available events to the relevant test transaction rather than counting all incomplete purchases as payment errors.

Turn the finding into a change you can check

For the example, the next design is: the affected address field is reachable, the error is described there in clear text, and other entries are retained. W3C explains that an automatically detected input error must be identified and described in text. The proposed change is not a complete accessibility assessment.

After implementation, repeat the same task and then review the trend in the consistently defined funnel. Record other changes too, such as to the offer or traffic sources. A successful task trial demonstrates the usability of this workflow; it does not establish a general revenue effect.

Sources

• Shopify — Behavior reports

• Shopify — Troubleshooting checkout payment issues

• W3C — Error Identification