Consent Mode v2: why your numbers are too low and what that does to Smart Bidding

There is a number almost every store knows and one almost nobody knows. The known one is the rejection rate of the cookie banner. The unknown one is what that rate does to the bidding algorithm. Because Google Ads does not treat an order without consent as missing information. It treats it as a click that did not buy.

That is the core of the problem, and it is overlooked in most discussions about consent mode. People talk about reporting, about missing numbers in the dashboard. That is annoying but not expensive. It gets expensive when Smart Bidding learns from a distorted data foundation and systematically pushes budget to where the consent rate is high, instead of where the margin is high.

What happens inside the gap

Picture a hundred orders that came through a Google ad. Some arrive with consent and are measured normally. Some arrive without it. If consent mode is set up correctly, the browser still sends an anonymous ping, from which Google models a share of those orders. The rest disappears.

Modelling is better than its reputation but has two conditions. First, Google needs enough data per day to model at all. Small accounts regularly fall below that threshold and then get nothing. Second, Google only models what it sees as a ping. Whoever loads gtag only after consent, the so-called basic mode, does not deliver those pings. Then the gap stays complete.

Why Smart Bidding suffers

The consent rate is not randomly distributed. It depends on device, browser, the age of the audience, the time of day and the country. Safari users decline differently from Chrome users, mobile visitors differently from desktop, Austria differently from Germany. The algorithm sees different conversion rates in all of these segments and draws a simple conclusion: where less is measured, less is bought. So it bids less there.

The result is a distortion that reinforces itself. Segments with low consent get fewer bids, so fewer clicks, so even fewer measured conversions. After a few weeks the algorithm has locked onto the audience that most readily clicks accept all. That is not necessarily the audience that brings the most margin.

The four levers, in this order

1. 1

   Measure and lift the consent rate

   Before touching any technology: do you know how many visitors consent? The number sits in your consent tool. Declining must be as easy as accepting, that is not up for debate. But wording, order and timing of the banner move the rate by double-digit points without becoming shady.

2. 2

   Consent Mode v2 in advanced mode

   gtag loads before consent, with all consent types set to denied. It then sets no cookies and sends no identifiers, but it sends anonymous pings. Google models from those. Check in Tag Assistant that the pings really go out before the banner is clicked.

3. 3

   Enhanced Conversions

   At purchase the email address is hashed and passed to Google, where it is matched against signed-in users. That closes part of the gap cookies could never close, and in Shopify, Shopware and WooCommerce it is usually a single setting.

4. 4

   Server-side tagging

   A tagging server on your own domain sets first-party cookies that Safari does not delete after seven days and passes conversions server-side. The most involved step with the biggest effect on data quality. Worth it from a budget where one percentage point of measurement accuracy is worth four figures.

What you have to do with your targets

The last point is almost always forgotten. If Google sees only 70 out of 100 orders, then a target ROAS of 400 percent is really a target ROAS of 570 percent. The algorithm tries to hit a goal that was defined on complete data, using data that is incomplete. It will bid too cautiously and leave reach on the table that would have been profitable.

That is why every tracking change comes with an adjustment of targets. After each of the four steps above the measured conversion rate rises, and the target has to follow, otherwise you suddenly spend far more than planned. The reverse also holds: as long as the gap exists, targets may be softer than the margin would actually allow.

If you are not sure where your account stands: the quick test above needs no agency and no technology, only two numbers. Whoever then sees a gap of a quarter or more has found the biggest lever in the account, and it is not in the campaigns.