Data Manager Conversion Diagnostics: Find the Missing Step

Find the point where the numbers stop agreeing

Google Ads reports fewer qualified leads. Before changing a bid target, compare that result with the CRM's actual qualified leads for the relevant period. If both fell, the business may genuinely be receiving fewer good leads. If the CRM is stable and the imported outcome is missing, investigate the delivery path.

In a hypothetical account, the CRM contains 42 newly qualified leads, but the latest import selected only 27. That is a much more useful clue than “conversions down 36%.” The missing 15 records give you something to inspect: their qualification times, source table, required fields and selection criteria.

Do not force the two systems to match by changing a reporting definition. Uploaded events, attributed conversions and total CRM outcomes are different populations. First explain the gap that the evidence can actually support.

Open the diagnostic that matches the failure

Tag Diagnostics examines the Google tag. Connection diagnostics examines connected sources and imports. The multi-source conversions diagnostic is a separate beta for setups combining tag and additional event data. Availability is account-dependent.

For a connection problem, open Data Manager and inspect the connection's status and details. Google distinguishes data that is still importing with issues from urgent failures that prevent imports. The underlying diagnostic matters more than the color: it may name rejected credentials, a missing source, a missing mapped column, or a failed run.

An enabled conversion action only describes its configuration. It does not demonstrate that a browser event fired or that today's CRM export completed. If the problem is on the website, reproduce the event there rather than repeatedly reconnecting a healthy data source.

Work the missing batch, not the whole account

Return to the 42-lead example. Suppose all 15 missing records qualified after 18:00. A source export that stops at 18:00 is a plausible explanation. So is a timezone mismatch in a filter. Neither is established until you inspect the source query and a few affected timestamps.

Compare the last successful run with the expected source contents. Check whether the failed rows share a field format or a recent schema change. Google documents timestamp transformations with a fallback timezone; missing timezone information without a fallback can stop rows importing. Transformation documentation can help distinguish a formatting failure from a selection mistake.

Once the cause is known, use the supported recovery path for that connection and event type. Do not resend an entire historical file just because some records are missing. Establish what was accepted and what identifiers prevent duplication before retrying anything.

Then inspect the result of the repair. A clean configuration screen is not enough: did the missing records pass the relevant import stage? Were any still rejected? Allow for the difference between a completed import and later attributed campaign reporting.

Multi-source conversions need an additional check

In the multi-source beta, the tag and the additional source must describe the same transaction consistently. Google's diagnostics can flag missing transaction IDs, suspected overcounting or large value discrepancies. An order reference that differs between the two sources can undermine deduplication even when both sources are sending data.

Google's current beta documentation describes a 14-day period in which additional-source data is used for reporting and diagnostics but not bidding; existing tag-source conversions remain biddable. It also says the additional data becomes biddable afterward regardless of diagnostic alerts. Do not assume an unresolved warning automatically prevents that transition. Read the trial-period details.

That deadline makes a real value or duplication error worth addressing promptly. It does not mean every account needs multi-source conversions, or that every ordinary Data Manager import has this trial period.

Finally, inspect which conversion goals the affected campaign actually uses. A broken secondary reporting event and a missing outcome that drives bidding have different consequences. Decide on a campaign response from that dependency and the size of the problem—not a blanket rule that all conversion movement requires pausing spend.