In this guide
Your campaign report shows 1,000 clicks. Your analytics report shows 820 sessions. Start by checking what each report counts. A click and a session are different events, so these illustrative totals do not establish a tracking fault on their own.
A useful investigation separates differences in definitions from differences in delivery or measurement. Work through the checks in order and keep the exports you compared.
Line up the reports
Before calculating a discrepancy, confirm these settings:
| Check | What to align |
|---|---|
| Scope | Campaign, placement, creative, account, and any included or excluded traffic. |
| Time | Start and end dates, time zone, and whether the current day is complete. |
| Metric | The event being counted: served impressions, clicks, sessions, purchases, or another event. |
| Cost | Currency, media cost, and whether fees are included. |
| Processing | Export time, reporting delay, filtering, and whether figures can still change. |
| Attribution | For conversions, which interactions receive credit and over what window. |
Attribution is the method used to assign credit for an outcome to advertising interactions. Two conversion reports using different attribution rules may answer different questions even when both are working correctly.
Calculate a comparable gap
Choose a reference report and state it explicitly. For this example, the signed difference is (Report B − Report A) ÷ Report A × 100. A negative result means B is lower than A. This convention is useful for reconciliation; it is not a universal billing rule.
The following sample compares the same impression metric after aligning dates, time zone, filters, and export timing:
| Placement | Report A | Report B | B compared with A |
|---|---|---|---|
| Homepage | 60,000 | 59,400 | −1.0% |
| Article pages | 40,000 | 32,600 | −18.5% |
| Total | 100,000 | 92,000 | −8.0% |
The larger gap is on article pages. Start there: compare devices, creative versions, and the dates when the gap appeared. Investigating the total alone hides that pattern.
If the reference count is zero, a percentage using that denominator is undefined. Report the counts. When combining rows, add the counts first and calculate the total difference from those sums.
Find where the event can be lost or counted differently
An ad click happens before the destination finishes loading. A visitor can leave before the analytics code runs. Redirects, consent choices, browser restrictions, and filtering can also affect what each system records. Google's clicks and sessions troubleshooting guide gives examples for Google Ads and Analytics.
For two impression reports, identify each system's counting point. One may record an earlier stage in delivery than the other. Read the relevant metric definitions before expecting the totals to match.
Look for changes in the pattern. Did the difference begin after a new creative, a landing-page update, or a change to consent handling? Record the timing and test the suspected cause. Timing is a lead, not proof of causation.
Send an escalation someone can investigate
Copy this structure and replace the sample details. Keep credentials and user-level data out of the message.
Issue: Report B is 18.5% below Report A for article-page impressions.
Scope: Campaign and placement IDs; affected dates and reporting time zone.
Reports: Metric names, filters, and export times for both systems.
Pattern: The gap is concentrated on article pages; homepage delivery differs by 1%.
Checks completed: Definitions and filters aligned; affected creative versions identified.
Request: Confirm the counting rules and investigate the attached sample rows.
Attachments: Two exports and a small comparison sheet with sensitive information removed.
Agree investigation thresholds with the parties involved. A fixed percentage is not a substitute for the campaign agreement, the size of the gap, or its financial impact.
Finish with what the evidence supports. If the cause is still unknown, state what was ruled out and what needs to be checked next.
AdOpsNow Editorial
Examples use illustrative data. See our editorial standards or report a correction.