Prepare data without double-counting records

Workroomly Editorial
Begin the lesson ↓

Share this resource

Copy the link to share this page with your team.

Choose the right row meaning and time boundary before calculating a report.

Prepare data without double-counting records

A chart can look plausible while counting the wrong thing. Decide what one row represents and use stable identities to connect related data.

Start with the right grain

If you are counting tasks, one analytical row should represent one task. A task with three review events should not become three tasks simply because the records were joined.

A join connects related records using a key. Use the task or item ID, not a title that may be reused or changed.

Prepare the records

1. Keep the original source unchanged and make an analysis copy. Note which source version and cutoff the report should use.

2. Check unique identities. Remove an exact duplicate event from the analytical count, but resolve conflicting versions of the same event with the source owner rather than choosing whichever looks better.

3. Apply the time rule. An event that occurred after the report’s cutoff does not belong in that earlier snapshot simply because it appears in the file you received.

4. If your measure uses the latest review, select the latest valid review at or before the cutoff for each item, then apply the success rule. Do not fall back to an earlier approval when the latest review requests changes. Join that latest review to the item list. Keep items with no review—they may be the exceptions you need to see.

5. Apply the intended population rules, such as active items due within a period. Keep due date, created date and review-event time distinct.

6. Compare the included identities with the source, then calculate totals. Equal row counts can hide a missing item and a duplicated one.

Use a simple check

If nine eligible items contain three qualifying results, the rate is 3/9. If one item appears twice, fix the identity problem before presenting the percentage.

Your result is a clearly defined analytical set. This preparation does not assume a particular Workroomly query language or automatic deduplication feature; use the supported reporting route for your actual source.

Continue learning

In this series: Build useful business reports — guide 3 of 8.

Turn what you learned into the way the team works.

Mark this lesson complete once the workflow exists in your workspace—not simply when you reach the bottom.