How to check customer interviews against product data

When customers say an approval step slowed setup, check the setup records before deciding to change that step. The records can help you choose where to investigate. They cannot, by themselves, explain why someone waited.

Here is a hypothetical example. You make a product that requires an initial data import. Several customers say they waited for a manager's approval before importing. You want to know whether this is a problem to work on next.

Compare the same task over the same period

Define the behavior you want to check: did each new account complete its first successful import within seven days of account creation? A login is not an import, and activity elsewhere in the customer's company may belong to a different account.

Use accounts that have all had the full seven days. Leave out every newer account, including those that already finished. Keeping recent successes while excluding recent unfinished accounts would make the result look better than it is. Seven days is the question in this example; choose a period that makes sense for your product.

Check that your records reliably capture account creation and the first successful import. A missing import event might be a tracking gap. Until you resolve that, it is not evidence that the customer failed to finish.

Mixpanel's funnel documentation explains counting a sequence of events within a defined period. The same care with the task, account and clock matters whichever reporting tool you use.

Do you actually know who needed approval?

Your interview notes may establish this for the people you spoke with. They do not label every other account in the dashboard.

Compare accounts that required approval with accounts that did not only if you have reliable information for both groups. Leave unknown cases separate. If approval requirements were never recorded consistently, the dashboard cannot yet answer that comparison.

Even if the approval group finished less often within the week, it might also have had more complicated imports. The difference would give you something to investigate, not establish that removing approval would help. Microsoft's product-research example distinguishes monitoring the current experience from testing a proposed change.

Give the next conversation a specific question

Look for an account you know required approval but still finished within the week. Start with its existing notes. If they leave the sequence unclear, someone who handled that account's setup could explain it.

Ask: “Walk me through what happened between creating the account and the first successful import.” Then ask where approval fitted into that sequence. Do not assume it caused a delay.

That conversation might reveal how this account got through setup. It would not show how common that experience is or settle whether approval slows other accounts. A general industry expert cannot fill in this particular customer's history.

This is where interviews and product records help each other: the records give you a more specific person and question to follow up on. Mixpanel's practitioner guide describes choosing between the methods based on what remains unanswered.

Check your existing notes and available customer contacts first. If you still need another person's experience, use the expert-selection guide to work out whose experience would answer the question before paying for another conversation.