Google is about to credit conversions 180 days after the click

Google announced a predictive metric at Marketing Live in May 2026 that credits conversions expected to happen up to 180 days after an interaction. It is called Qualified Future Conversions, it runs on Gemini, and it is in restricted pilot with a wider beta expected late in 2026.

The number underneath it is the interesting part. Google says only 40% of Demand Gen conversions land in the first 30 days. Which means that for six years, everyone reporting a 30-day window on upper-funnel campaigns has been reading roughly two fifths of the story and making budget decisions on it.

What it actually does

A conventional conversion column counts things that happened. This one counts things a model expects to happen, based on the behaviour it has observed so far in the customer journey.

That is a genuine step, not a relabelling. Long consideration cycles have always been invisible to in-platform reporting, which is why brand and upper-funnel budgets get cut first in a downturn. They look like they do not work because the payoff sits outside the reporting window.

Your historical numbers are about to look wrong

Here is the practical problem that will land on your desk before the feature does.

When predicted conversions appear in your account, your cost per acquisition on affected campaigns will drop, sometimes sharply. Nothing will have changed in the market. The measurement changed.

If you report to a board on cost per acquisition trend, you now have a break in the series that has nothing to do with performance. If your bonus structure or agency contract references a cost per acquisition target, that target was set against a different definition.

Three things to do now, before it arrives.

1. 1.Take a baseline. Freeze your current cost per acquisition and conversion volume by campaign for the last twelve months, exported, in a file you own. Once the metric changes, you cannot recreate the old view.
2. 2.Tell finance early. The sentence you need is that a portion of reported conversions will be predicted rather than recorded, and predicted conversions cannot be reconciled to invoices. Say it before the first report, not after.
3. 3.Keep an actuals view. Whatever else you report, keep one view that counts only conversions that have actually occurred, sourced from your CRM. It is the only number that ties to money in the bank.

Prediction in the bidder is not new, prediction in the report is

Automated bidding has always been predictive. Target CPA is a forecast. Nobody objected, because the prediction stayed inside the machine and the reported conversions stayed real.

Putting a prediction into the reporting layer is a different thing. It means a number in your dashboard is a model output about the future, sitting in the same column as a number that describes the past. Anyone reading the report needs to know which is which, and by default they will not.

Push for a separate column or a clear segment. If you cannot get one, annotate every report you send.

Where it will genuinely help

Long sales cycles. If you sell professional services, equipment, construction, education or anything with a 60 to 180 day consideration period, the gap between click and contract has always been the reason your upper funnel could not defend its budget.

Two campaigns can look identical at 30 days and be worth completely different amounts at 180. Until now the only way to see that was to build the pipeline attribution yourself, in your CRM, with campaign data stitched back in. Most teams never did, because it is genuinely hard.

The honest caution

A prediction is a model, and models are wrong in patterned ways. It will likely be better calibrated for high-volume advertisers with clean conversion data and worse for everyone else. It may be optimistic in the first version, since the incentive structure points that way.

The defence is not to refuse the metric. It is to check it. Six months after you turn it on, compare what was predicted against what your CRM says actually closed. If it is close, use it. If it is not, you now have the one thing nobody else has, which is a calibration factor for your own account.

The teams that come out ahead here are the ones already exporting their own conversion truth every month, because they will be able to grade the prediction instead of taking it on faith.