Skip to main content

When metrics don't match expected numbers

Troubleshoot differences between Zoë's numbers and another dashboard, spreadsheet, or reporting tool.

When metrics do not match expected numbers

If Zoë's numbers do not match a spreadsheet, dashboard, or another reporting tool, the cause is usually one of these:

  • The date range or timezone is different

  • The metric definition is different

  • A filter is applied in one place but not the other

  • Duplicate rows or joins are inflating the result

  • The data model uses a different table or field than expected

  • The source data has changed since the comparison report was created

Step 1: Check the date range and timezone

Start by making the comparison as specific as possible.

Example questions:

  • “Show revenue from January 1 to January 31, 2024.”

  • “Show revenue for January 2024 by day.”

  • “What timezone is this result using?”

Small date or timezone differences are one of the most common reasons numbers do not match.

Step 2: Check the metric definition

Ask Zoë how the metric is calculated, or ask to see the SQL if your role allows SQL visibility.

Example questions:

  • “How is revenue calculated?”

  • “What fields are used for this metric?”

  • “Show me the SQL for this result.”

Compare that definition to the spreadsheet, dashboard, or external report.

Step 3: Check filters and segments

Confirm both results use the same filters.

Common differences include:

  • Region

  • Product

  • Customer segment

  • Order status

  • Paid vs unpaid orders

  • Test or internal accounts

  • Currency

If one result has a filter that the other does not, the numbers will not match.

Step 4: Check the underlying row counts

Before comparing calculated metrics, compare the raw data volume.

Example questions:

  • “How many orders are included in this result?”

  • “How many rows are in the orders table for this date range?”

  • “Show order count by status for this period.”

If the row counts differ, investigate the table, joins, filters, or access rules before troubleshooting the metric formula.

Step 5: Check joins and duplicates

If a metric is higher than expected, a join may be duplicating rows.

Ask a data model owner to confirm the joins, identifiers, and primary keys used by the metric.

When to contact support

Contact support if the metric still looks wrong after checking date range, definition, filters, row counts, and joins.

Include:

  • Workspace name

  • The exact question asked

  • The result Zoë returned

  • The number you expected

  • The report, dashboard, or spreadsheet you are comparing against

  • Screenshots or SQL, if available

Did this answer your question?