Sep 9, 2026
Google Ads Now Ignores Offline Conversions Uploaded After 7 Days
Google has introduced a limit that is easy to miss because nothing visibly breaks: conversions uploaded more than seven days after the event no longer count towards data-driven attribution. They still appear in your reports, the numbers in the interface stay the same. What changes is invisible: the algorithm stops learning from them. For any business with a long sales cycle, that means some of the most valuable data you have is now going nowhere.
0 days
maximum upload delay before the data is ignored
0%
of those conversions still show up in reports
0
effect on attribution and on algorithm training
0-50
conversions a bidding strategy needs to run steadily
If you import conversions from your CRM once a month, or export deals only after they close, that data no longer influences data-driven attribution. There is one fix: upload more often than every seven days, and record intermediate stages rather than only the payment. The ideal setup is an automated daily import instead of manual exports.
What changed
Offline conversions are the mechanism that feeds back into Google Ads what happened away from the site: a call turned into a deal, an enquiry reached payment, a client signed. You export that data from your CRM together with the click identifier, and the system can see which ads produce actual revenue rather than just enquiries.
That mechanism now has a time threshold. Data uploaded more than seven days after the event itself stays in reports, but the data-driven attribution model disregards it.
| Scenario | Shows in reports | Affects attribution |
|---|---|---|
| Deal closed on the 1st, uploaded on the 3rd | Yes | Yes |
| Deal closed on the 1st, uploaded on the 9th | Yes | No |
| Monthly export routine | Yes | Almost never |
| Automated daily import | Yes | Yes |
Because on the surface nothing happens. You still see the conversions, the count does not drop, no alert fires. But automated bidding stops treating that data as signal and reverts to optimising for whatever arrives quickly: cheap enquiries instead of expensive customers. You will notice in two or three months, when lead volume is up and sales are down.
Who this hits hardest
- B2B and complex services. Sales cycles from two weeks to several months. This is exactly where offline conversions were the only way to show the system where real value sat.
- Healthcare and clinics. Enquiry today, appointment next week, payment later still. Uploading after the visit puts almost everything past the threshold.
- Real estate and high-value purchases. Months can pass between first contact and closing, so intermediate stages have to be recorded rather than only the final one.
- Anyone exporting manually. The most common routine is once a month before the report. That now has almost no value for optimisation.
- Education and courses. Someone enquires, thinks for two weeks, then pays. A textbook case of the final event falling outside seven days.
What to do
- Check how often your import actually runs. Tools and settings, then uploads. If the last date is last month, you are already losing data.
- Move to automated daily uploads. Via the Google Ads API, a CRM connector, or a scheduled import from Google Sheets. Manual exports are no longer an option.
- Record intermediate stages, not just payment. "Qualified lead", "meeting booked", "quote issued" can all be sent as separate conversions with different values. They happen sooner and fit inside seven days.
- Assign a value to each stage. If a qualified lead closes 20% of the time at an average deal size of $1,500, that conversion is worth $300. The system then optimises against it correctly.
- Confirm the click identifier is being stored. Without it the upload fails outright. Its lifetime is limited, so the CRM must capture it the moment the enquiry arrives.
- Do not switch off your old exports. For reporting they work exactly as before. Just stop relying on them when configuring bidding.
If there is no proper CRM integration, the cheapest route is a Google Sheet on an import schedule. A manager records closed deals in the sheet on the day they close, and Google Ads pulls it automatically every day. It takes five minutes to set up and solves the seven-day problem with no development work at all.
How to restructure around the limit
The logic is simple: the closer an event sits to the click, the more reliably it reaches the algorithm. So a long customer journey has to be broken into several short ones.
| Stage | When it happens | What to send |
|---|---|---|
| Form submission | immediately | Base conversion with a small value |
| Contact and qualification | 1-2 days | Separate conversion, higher value |
| Meeting or quote | 3-7 days | Another conversion, the primary signal for the algorithm |
| Payment | weeks or months | Send it for reporting; it will not affect attribution |
This approach carries a second benefit unrelated to the limit itself. The algorithm needs 30-50 conversions to run steadily, and a B2B account may close five deals a month. Intermediate stages supply enough data volume in exactly the accounts where final events are always scarce.
How to check whether you are already losing data
Before rebuilding anything, confirm the problem is actually yours. Three checks take about fifteen minutes.
- The date of your last upload. In the account: Tools, then Uploads. Look at the date column. If the gap between events and uploads is consistently over a week, you are in the risk zone.
- The gap between conversion date and import date. Take your most recent export file and calculate that gap for twenty rows. An average above seven days means most of your data is not feeding optimisation.
- The attribution model on your conversion actions. If it is set to last click, the limit does not affect you at all yet, because the data-driven model is not in use. But moving to it in this state is unwise until the import is fixed.
If your account manager exports deals "when they get round to it" rather than on a schedule, assume the data is already inert. Automating this delivers more than any bid adjustment: the algorithm gets a truthful picture instead of inferring one.
What this costs in money
An abstract "loss of signal" sounds harmless, so here it is in numbers. A B2B services company spends $1,500 a month, gets 120 enquiries, and 12 of them become deals at an average value of $1,200.
When the algorithm can only see enquiries, it hunts for the cheapest ones. Cost per enquiry falls from $12.50 to $10, volume rises to 150, and the report reads like a 25% improvement. But within those 150 the share of qualified enquiries drops, because the system is now optimising for people who part with a phone number easily and do not buy. Deals fall to 9 instead of 12.
In money: three fewer deals a month is $3,600 of revenue. So saving $375 on cost per enquiry cost nearly ten times that. This is what a disconnected signal costs, and none of it appears in the reports.
Common mistakes during the switch
- Giving every stage the same value. The system then optimises for the cheapest and most frequent one, which is the plain enquiry.
- Setting every stage as a primary conversion at once. One event should be primary; the rest belong in secondary, observation-only.
- Forgetting refunds and cancellations. If a deal falls through, the conversion needs adjusting, otherwise the algorithm learns to find people who back out.
- Uploading without the click identifier. The data is simply rejected, and you will only find out in the error log.
Frequently asked questions
Will my conversions disappear from reports?
No. They display exactly as before and the counts do not change. What is lost is their influence on the attribution model and on automated bidding.
Are the seven days counted from the click or the event?
From the conversion event itself. If a deal closes on Monday, you have until the following Monday to upload it.
Does this apply to enhanced conversions?
Enhanced conversions for leads use the same upload mechanism, so the rule reaches them too. Conversions recorded on the site at the moment of action are unaffected entirely.
What if deals take three months to close?
Optimise towards an intermediate stage that happens quickly, and use the final payment for reporting and for checking whether the two have diverged. If qualified leads rise while deals do not, the chosen stage was the wrong one.
Does this affect short-cycle campaigns?
Barely. E-commerce, delivery, appliance repair: the conversion is recorded on the site immediately and no upload is involved.
This change demands no emergency action today, but a quarter of quietly running the old routine leaves you with campaigns optimised for enquiries rather than revenue. If you want us to review how your conversion import is set up and rebuild it around the new limit, talk to our team. If you would rather understand how account analytics should be structured in the first place, start here: how to launch Google Ads from scratch.
Want results like this for your brand?
Get a free strategy call and a tailored proposal within one business day.
Get a Free Proposal