A new number has started showing up in some Google Ads accounts: “5% additional conversions reported within the past 14 days.” That is the Data Strength Uplift, the metric Google announced on September 10, 2026. Let me be blunt: in nine reports out of ten, this number is going to be misread. A busy reader sees “+5% conversions,” concludes “+5% sales,” and picks the wrong metric. What the Data Strength Uplift counts is recovered measurement, not created demand. The distinction sounds academic. It changes everything you are allowed to write in a client report.
This guide does the work that neither Google nor the news briefs will do: cleanly separate recovered conversions from incremental sales, open the announcement’s footnotes one by one, and give you the procedure to sanity-check the number before you put it in front of anyone.
What Google announced on September 10
The announcement is signed by Nipoon Malhotra, VP of Ads Analytics at Google. It stacks several pieces: Data Manager integrating directly into Google Analytics and DV360, a Data Manager API that becomes “universal” (it adopts the IAB Tech Lab’s ECAPI standard), enhanced conversions extended to GA and DV360, global general availability of Meridian GeoX, and, in the middle of all that, the Data Strength Uplift.
On that last point, the rollout is in progress. The metric is not in every account as I write this. The right reflex is not to take my word for it, it is to go look in your own interface. On the universal Data Manager API and what it means for your existing pipelines, I have covered elsewhere whether you should migrate from the Measurement Protocol to the Data Manager API: the short answer is “not in a panic.”
What the Data Strength Uplift actually counts
The Data Strength Uplift estimates the number of additional conversions recovered thanks to your first-party data setup. In plain terms: without enhanced conversions, without a tag gateway, without CRM matching, Google loses a share of the conversions that did actually happen. A user bought something, but the signal never made it back (blocked cookie, cross-device journey, Safari’s ITP, partial consent). When you improve your setup, Google recovers some of those lost conversions and tells you: “look, your configuration let us reconstruct X% more conversions.”
The example screen Google provides is explicit: a small box in the corner of the interface reading “5% additional conversions reported within the past 14 days.” The word that matters is reported. These are conversions being reported, not conversions being generated.
That is a good thing, and I am not going to trash it. Recovering lost measurement is exactly what I do all day for clients, and it is real value: better data feeds Smart Bidding, which steers bids better. But the metric’s value stops at what it measures. The problem is not the metric, it is the use people will make of it.
What it does not count: the section not to skip
Here is the trap, and it is structural. Recovering measurement is not creating demand.
Say you sold 1,000 products this month. Before your new setup, Google only saw 850 of them (the rest got lost in the plumbing of tracking). After you turn on enhanced conversions, Google sees 950. So your Data Strength Uplift shows a nice conversion uplift. How many extra sales did that setup produce? Zero. You still sold 1,000. You simply count 100 more. The sale already existed; it is the measurement report that moved closer to reality.
That is why the same announcement contains Meridian GeoX, and it is no accident. Google frames its measurement in three layers: a data foundation (where the Data Strength Uplift lives), multiple signals, and causal proof. The question “did my marketing create additional sales” is not answered with recovered data, it is answered with an experiment. That is exactly GeoX’s job: a geo test with a control group that isolates what your campaigns actually caused. If the topic interests you, I wrote a full guide on preparing a geo incrementality test with GeoX and GA4. Just keep the hierarchy in mind: the Data Strength Uplift answers “am I measuring well,” GeoX answers “does it sell more.” They are not synonyms, and one level up, the Meridian MMM arbitrates budget across channels.
The footnotes you have to read
The announcement comes with four punchy numbers. The body text presents them as general averages. The footnotes tell a more nuanced story. I opened each one on the primary source. Here is what they actually say.
| Headline number | What the body text implies | What the footnote actually says |
|---|---|---|
| +14% conversions via Google tag gateway | An average gain across all verticals | Google Internal Data, Global, Finance sector, Jul-Dec 2024 vs Jan-June 2025. Scoped to a single vertical, on 2024/2025 data for a 2026 announcement. |
| +20% on Demand Gen campaigns | A solid, generalizable gain | Google Data, Global, Performance, June 3 to June 17, 2026. A 14-day window. |
| +11% Search conversions with enhanced conversions | Enhanced conversions vs nothing | Enhanced conversions vs standard conversion imports, January 1 to 14, 2026. A 14-day window, against an already-decent baseline. |
| +26% incremental ROAS (offline + app connected) | A gain that applies to every account | Google Data, Global, Measurement, April 2025-April 2026, on Search campaigns bidding to conversion value. Not lead gen on CPA, not other networks. |
Three things to note, without any trial of intent. First, these are Google internal figures, not independently auditable, produced by the platform that directly benefits from you sharing your data. That is not an accusation, it is a fact to keep in mind when citing a number. Second, the scope is almost always narrower than the headline suggests: one vertical (Finance), one campaign type, a two-week window. Third, 14-day windows over ultra-short periods are fragile: they also capture seasonality and noise.
If you want to understand where the tag gateway’s 14% concretely comes from, meaning what that brick actually does, I compared the Google tag gateway and server-side GTM: two ways to take back control of collection, with very different implications.
How to sanity-check the number before you publish it
This is where the real value lies for a practitioner. A non-auditable number handed to you by the platform is not a reason to reject it; it is a reason to cross-check it before it goes into a report. Here is the procedure I apply.
1. A before/after on your own data. Take your GA4 export in BigQuery, or your sGTM data if you log server-side, and look at conversion volume around the exact date you turned on your setup. Google’s uplift should show up, at least in the same order of magnitude, in your own count. If it does not show up at all, something is off somewhere, and you want to know before your client does.
2. Check that no other change is polluting the window. This is the step everyone skips and it is the one that voids conclusions. Did you change your conversion window during the period? If so, your count moves for a reason that has nothing to do with data strength (see how to choose your GA4 conversion window). Did the attribution model shift? The GA4 attribution changes of 2026 moved numbers without anyone touching a thing. Were URL parameters stripped along the way? That is a classic cause of false uplift, in the same family as the gad_source trap that counts Ads traffic as organic. Until you have ruled these out, you cannot attribute the variation to your setup.
3. If the real question is “does it sell more,” switch tools. No amount of cross-checking recovered conversions will answer a causality question. At that point, you move to a GeoX test. It is the only honest way to tell leadership “yes, this budget created sales.”
An action plan to actually raise your data strength
If you want this number to climb for real, priority order matters. Here is how I rank it in the field, from highest to most marginal return.
- Tag gateway or server-side GTM. The brick that takes back control of first-party collection. Best effort-to-impact ratio, and incidentally the source of the announcement’s 14%.
- Enhanced conversions. The unified toggle since April 2026 made activation much simpler. Server-side, you typically recover 10 to 20% of lost conversions. For a B2B context with forms, look instead at enhanced conversions for leads in server-side GTM.
- Data Manager and CRM. The central hub that unifies and activates your first-party data. It is the brick gaining weight with this announcement; I wrote the full guide.
- Offline data. The last link, the one feeding that 26% iROAS figure, provided you are bidding to conversion value.
Decision box: what to say, and to whom
When should you cite the metric? When you are talking about measurement quality and Smart Bidding steering. There, a data strength uplift is a legitimate, verifiable argument.
When should you keep quiet about it? When the conversation is about growth, incrementality, or real ROI. In that context, pulling out a Data Strength Uplift is at best off-topic, at worst misleading.
And if a client shows you the number before you have seen it? Do not tear it down, explain it. “This number is good news: your setup is reporting your conversions better. It does not say you sold more, it says you measure better. To find out whether we sold more, here is what we put in place.” That sentence is what sets you apart from the agency that copy-pasted Google’s press release.
The Data Strength Uplift is a good indicator of your tracking’s health. It is not a business performance indicator. Treat it for what it is, cross-check it before you publish it, and keep GeoX for questions of causality. Your reporting will gain credibility, and that is the only thing that counts when you have to defend a budget.