Your Google Ads account shows conversions in two places. What most advertisers don't know is that those two numbers are drawn from two completely separate engines — and one of them silently drops a portion of your CRM data before it ever touches your bids.
The constraint was spotted by PPC consultant Hana Kobzova and brought into wider circulation by Adriaan Dekker, a Google Ads specialist, in a LinkedIn post on August 22, 2026. The rule: any offline conversion uploaded more than 7 days after the underlying event occurred is excluded from Data-Driven Attribution (DDA) calculations. It still appears in your standard conversion column. DDA ignores it. Smart Bidding runs on DDA.
Two numbers in the same account
Google Ads surfaces conversion data in standard reporting and in attribution reports. These have always differed somewhat — different lookback windows, different modeling logic. But they generally pointed in the same direction. This 7-day rule creates a harder, structural divergence.
Standard conversions count everything you upload, regardless of when you upload it. If a customer converted on August 25 and your CRM batch export ran on September 2, that conversion shows up in your "Conversions" column — a 8-day gap. Standard reporting records it.
DDA never saw it. The attribution engine only processes conversions that arrived within 7 days of the event date. Anything beyond that threshold is bypassed. The conversion counts in one place and is invisible in another — both columns sitting in the same account, drawn from the same event data.
There's no alert. No flag in the interface. The account simply continues reporting and bidding as if both numbers are consistent.
Why this matters for Smart Bidding
Data-Driven Attribution isn't just a reporting layer. It's the primary input for Target ROAS and Target CPA bidding. When DDA's conversion history is incomplete, Smart Bidding allocates budget and sets bids based on a partial picture of your actual sales data.
In practice, this means campaigns that drive late-uploading conversions appear to the algorithm as lower-performing than they really are. Over time, Smart Bidding moves budget away from those campaigns, keywords, or audiences — not because they're underperforming, but because the attribution model can't see their results.
This isn't a reporting discrepancy you can shrug off. It's a bidding miscalibration that compounds over weeks and quarters as the algorithm reinforces its own blind spots.
Who's most exposed
Weekly CRM export schedules create the clearest risk. If your batch job runs every seven days, conversions recorded at the start of that cycle are already approaching the threshold by the time they upload. Any variance in timing — a delayed export, a weekend job failure — can push conversions past the cutoff.
Sales cycles longer than one week are structurally at risk. A customer who requests a demo on a Tuesday and confirms purchase the following Wednesday is converting ten days after first contact. That offline conversion, uploaded on any schedule, arrives after the 7-day window. DDA can't touch it.
Also at risk: businesses where the CRM-to-Ads connector has any latency built in. HubSpot's native Google Ads integration, Zapier workflows, and custom scripts all introduce their own timing variability. An export that nominally runs "daily" can still push late if a connector retries after a failure.
Lead gen accounts are especially exposed. High-ticket purchases, subscription trials that take time to confirm, and B2B inbound inquiries that convert through sales follow-up are all conversion types that regularly exceed the 7-day threshold.
What to check now
Start by pulling your offline conversion upload logs. Most CRM connectors surface this in their reporting — you want the timestamp of the upload, not the conversion date. Find the gap between event date and upload date for the last 30 days and see how many conversions are landing past 7 days.
If you're working with batch exports, move them to daily. This is the single most direct fix. A daily export on a stable schedule keeps your upload latency below 3–4 days for almost all conversions, providing enough buffer even if there's occasional job lag.
Check your attribution model setting across all conversion actions. If any of your key conversion actions are still on a non-DDA model, this specific issue won't affect them — but DDA is now the default for most accounts, and Google has been migrating holdouts throughout 2026.
The comparison that reveals the problem most clearly: take your CRM-reported revenue for the last 90 days and put it next to your Google Ads attributed conversion value for the same period. If your CRM number is meaningfully higher, and your model is DDA, late-uploaded conversions are the first variable to rule out.
If you want to see how offline conversion gaps are affecting your specific account's bidding logic, the free audit at Gromerce surfaces attribution mismatches that don't show up in your standard dashboard.
Two numbers in the same account don't have to disagree. They will — unless your upload timing is clean.
Sources: PPC.land, Adriaan Dekker via LinkedIn, Search Engine Land — September 2026
