Data Strength Uplift counts Google's gain, not yet yours

Over 20% is the figure at the top of Google's October Demand Gen update. Based on an internal analysis, Google says, advertisers who build their data strength with Google Tag Gateway see, on average, over 20% conversion uplift in Demand Gen campaigns.

That is an internal analysis and an average, in Google's own words, and it is the backdrop to the new metric the same update launches. Google says the Data Strength Uplift metric is launching in Google Ads, helping you understand the conversion gains driven by your data setup within the Google Ads UI. The update also says Data Manager is expanding to Display & Video 360 and Google Analytics, enabling you to connect and manage your first-party data, and that enhanced conversions will soon be available in Display & Video 360 and Google Analytics.

Its case study is an insurance platform that, Google says, implemented data strength to achieve a 16% increase in conversion volume and a 19% increase in overall conversion value.

The announcement does not say how the uplift metric is calculated, or what Google Tag Gateway does, and the Google documentation we keep does not cover either yet. So this post does not guess at either mechanism. What it can do is say what the pieces of a data setup change, from Google's own pages on them.

Here is the distinction the uplift figure does not draw for you. Google says modelled conversions use data that doesn't identify individual users to estimate conversions that Google is unable to observe directly, and that its modelling determines whether a Google ad interaction led to the online conversion. It doesn't determine whether or not a conversion happened. In the Conversions column, Google reports both modelled and observed conversions.

Enhanced conversions works on the same gap from the other side. Google says enhanced conversions sends hashed first party conversion data to Google, and, for enhanced conversions for web, that the data is then used to match your customers to Google Accounts which were signed in to when they engaged with one of your ads. In its note on cross-device conversions, Google says the share that can be recovered depends on the amount of observable data and how representative it is, and names the Google tag, consent mode and enhanced conversions as ways to improve on it.

Our reading, not Google's: the pieces covered here all work on the link between an ad and a sale your business was already recording somewhere. So a rise in the Conversions column after improving them is, at least in part, sales you already had now being credited to Google Ads. That is real and useful information. It is not, by itself, new revenue.

None of that makes the setup cosmetic. Google says that when it loses observation for certain cohorts of users, its automated bidding algorithms will need to make optimisation decisions based on incomplete data, resulting in biased learning, and that automated bidding may deprioritise those cohorts since they have a lower reported performance. Modelling, it says, corrects these biases in overall reporting so that automated bidding has access to more representative performance data.

Our reading: a bidder that cannot see a customer's conversion learns that the customer is worth less than they are. Better data can change where the money goes, and that is the part of an uplift that could show up in your sales. The only way to know whether it did is to look there.

This is Riibon's practice, not Google's. A recommendation we make about a live account cites an outcome measured outside the platform: bookings from the calendar, payments from the payment processor, deals from the CRM. The uplift metric gets the same treatment.

First, know which pieces you have. A Google tag on every page, not only the checkout, because Google says its products only collect data on pages where you've deployed the associated tags. Consent mode, and which version: Google says basic consent mode bases conversion modelling in Ads on a general model, while advanced provides an advertiser-specific model. Enhanced conversions, for web or for leads. And, if the sale happens off the site, an offline conversion import.

Second, wait before reading. Google says modelled conversions can take up to five days to fully process and stabilise in Google Ads reporting, and that conversion values are subject to retroactive increases for a period of several days. Consent mode has an uplift report of its own, separate from Data Strength Uplift: Google says that after at least seven full days, Google Ads and Google Analytics may have enough data to report it, and that the uplift numbers will only be shown when a slice's uplift meets the minimum data threshold.

Third, put the before and after next to your own count. If the Conversions column rose and the sales your business recorded over the same weeks did not, the gain is in measurement. If both rose, compare the two rises before crediting the setup, because nothing else in the account stops changing while you test.

Fourth, write down the date each piece went live. A Conversions column that changed what it can see halfway through a window cannot be compared with itself, and a change log is the only thing that will tell the next person why.