A successful offline conversion upload is not enough evidence to move budget between brands. In a first-hand investigation published on 5 November, StubGroup CEO John Horn describes Google Ads imports appearing in accounts that did not generate the originating interaction. His team also found gaps between successful upload rows and conversions visible in reporting. For an agency or multi-brand advertiser, the immediate task is to reconcile the evidence behind account-level results.
Separate the observation from a platform-wide conclusion
The reported case involved a multi-brand client with separate advertising accounts and conversion actions sharing the same name. Horn describes repeated acceptance of some identifiers across accounts, alongside other imports being rejected. This is a practitioner’s account of observed behavior, not proof that every advertiser has the same problem or that all offline imports are unreliable.
The effect on automated bidding remains unresolved in the investigation. The author reports assurances from Google Support but says he could not independently verify the bidding treatment. A CMO should therefore avoid both claims: that bidding is definitely damaged, or that there is no possible optimization concern. Record the uncertainty and investigate the accounts actually used for decisions.
Build a transaction-to-account evidence trail
Our recommended control is a reconciliation ledger linking each CRM event to its intended account, conversion action, identifier, event time, value and upload batch. Keep the original export and import response. Distinguish three questions: was the row readable, did it belong to this account and action, and how was the event eventually represented in reporting?
Start with a small sample of completed commercial events rather than a total from a dashboard. Trace each sample through the source system and upload destination. If several brands use the same action name, do not treat that name as sufficient proof of ownership. Check the intended account mapping and whether the underlying conversion event is genuinely shared.
Compare the same actions and time basis
Google’s import guidance recommends the “All conv. (by conv. time)” column for validation. Segment by conversion action and allow for processing and conversion delays before comparing. A CRM export organized by sale date and a standard advertising report organized by interaction date can disagree without demonstrating lost or duplicated sales.
Document the comparison window, time zone, action scope, counting settings and values. Investigate an apparent surplus separately from missing events. A successful file status should trigger the next validation step; it should not close the audit. Avoid repeatedly resending the entire history as an improvised response, because that obscures which changes explain the discrepancy.
Repair the routing before changing the budget
For account-specific events, review whether uploads can be partitioned by their intended destination. A shared manager-level conversion architecture may suit some businesses, but it is an operating-model choice rather than a quick cosmetic fix. Google warns that changing between account and manager tracking can change campaign goals, so review the goals before any migration.
Until the sample is reconciled, label affected account comparisons as provisional and postpone reallocations that depend on the disputed CPA or conversion value. Keep a named owner, a documented correction and a fresh validation sample. Resume normal allocation decisions when the commercial event, destination and report can be explained together; do not invent a universal discrepancy threshold or assume that changing an action name guarantees a fix.
Sources
- John Horn: Google Ads may count offline conversions in the wrong accounts
- Google Ads: Guidelines for importing offline conversions
- Google Ads: About cross-account conversion tracking
