Conversion tracking troubleshooting: missing, duplicate or wrong results

1. Find the first checkpoint that fails

Record the action time and time zone, intended event, source identifier, expected value and currency, and a non-sensitive test reference. Start from the application record: if the form or checkout failed, there should be no successful Lead or Purchase event.

Then inspect the installed integration and the platform’s event diagnostics. Only after receipt is confirmed should you investigate advertising attribution. Keep customer payloads and credentials out of shared screenshots.

On a narrow screen, scroll the table sideways to compare.

2. No events: check the trigger and destination

Confirm that the success trigger ran, not just the page-view tag. Check the exact source ID and integration environment; test, staging and production can point to different destinations. Verify that the deployed container or integration is the one you inspected.

Review browser diagnostics and permitted server delivery logs for blocked or rejected requests. Respect the visitor’s privacy choices: do not disable consent controls or blockers to manufacture production results. For a server response, inspect per-event validation and partial errors rather than treating an HTTP success code as final receipt.

• Verify the business success state before the tracking trigger.
• Match the deployed source ID to the source opened in the platform.
• Use the platform’s documented test mode; do not create artificial ad clicks.

3. Duplicates: trace every sender and retry

List each sender: direct code, tag manager, commerce plugin, browser Pixel and server integration. A second sender or a success-page reload may report the same action again. Give one business occurrence a stable reference and preserve it on retries; do not reuse that reference for unrelated orders.

Use the receiving platform’s exact deduplication contract. Similar names are not interchangeable, and a transaction reference is not automatically the field used to match browser and server events.

On a narrow screen, scroll the table sideways to compare.

4. Wrong value or currency: compare the same definition

Agree whether the conversion value represents a total order, net revenue or another documented amount. Compare the application record and event payload using that definition. Check decimal separators, major versus minor units, currency codes, refunds and any default value configured in the platform.

Do not silently convert USD, USDT or another currency into one another. One hundred dollars must not accidentally become ten thousand because an amount in cents was treated as dollars. Use the platform’s documented numeric format and verify the received fields with a permitted test.

5. Events received, but no conversions in the ad report

First confirm that you are viewing the correct account, source, conversion action, date range and time zone. Review the campaign’s selected goal, reporting column, counting configuration and attribution rules. A test event without an eligible advertising interaction may be received correctly without becoming an attributed advertising conversion.

Separate reporting delay from a rejected event. Inspect platform diagnostics and the relevant reporting definition before resending anything. For CRM or offline integrations, confirm the supported match identifiers and event time requirements, using only data you are authorized to share. Do not fabricate click IDs or replay customer events to increase a count.

6. Retest success, failure and retry

After one scoped correction, run a permitted success test, a failed action and a reload or retry of the same success. The intended success should be received once; failed actions should not become successful conversions. Keep a small, redacted evidence record of the implementation version and each result.

When the issue is platform attribution rather than delivery, record receipt separately and leave attribution unconfirmed until the relevant report supplies evidence. Do not mark the entire chain complete based only on a browser extension or a tag preview.

7. Hand over evidence without secrets

Share the platform, account or source reference through the established private support route, along with the symptom, expected result, time zone, redacted error and implementation owner. Note what has already been tested to avoid repeated changes by different people.

Never publish access tokens, API secrets, browser cookies, full customer payloads or payment details. AdsGorilla account support and a custom tracking implementation are separate scopes; confirm responsibility before requesting production changes.