Google Says Better Tracking Gets You 14% More Conversions. Counted, Not Sold.

Blog · September 27, 2026 · 11 min read

If you run Google Ads, a new number is about to appear next to your campaigns: Data Strength Uplift. Google describes it as the "additional conversions recovered by your first-party data setup". The example in Google's announcement shows 5% more conversions reported over 14 days.

The question for an Indian business spending ₹30,000 to ₹3,00,000 a month on Google is simple: does fixing your tracking bring more customers, or more conversions in a report? Mostly the second. One part of Google's package, importing lead quality, changes what you pay for. The rest changes what you see.

Every Google claim below comes from Google's announcement, its footnotes or its help pages. The independent evidence comes from two large field experiments, at eBay and at Facebook. Links are at the bottom.

01What did Google announce on 10 September 2026?

The post, by Nipoon Malhotra, Google's VP of Ads Analytics, Insights and Measurement, put Data Manager, Google's tool for connecting customer and offline data, inside Google Analytics and Display & Video 360. It added the Data Strength Uplift metric to Google Ads and upgraded Meridian, Google's open-source marketing mix model.

Four numbers carried the announcement. Each has a footnote at the bottom of Google's page, and the footnotes are where the story is.

Source: Google Ads & Commerce Blog, "Drive profitable growth with new data and measurement tools", 10 September 2026, footnotes 1 to 4, paraphrased. The last column is our reading.

Three things stand out. The 11% is not about the enhanced conversions most small businesses would switch on, the version for website purchases. It compares two ways of importing offline conversions. If you never upload leads or sales into Google Ads, that number is not about you.

The 14% compares the second half of 2024 with the first half of 2025. Google does not say whether the Finance label means the finance vertical or the internal team the data came from. Either way, advertiser budgets, seasonality and Google's own ad products all changed between those two periods, and a before-and-after comparison cannot separate any of that from the tag.

The 26% is the most interesting claim, because incremental ROAS is the right thing to measure: revenue the ads caused, not revenue they were near. But advertisers who connect offline data to Google are not a random sample. They are larger and have CRMs and measurement teams, and Google does not say how it separated the data connection from the kind of company that makes one.

Two of Google's four headline numbers come from two-week windows. A third compares two different half-years with no control group disclosed.

02What does Data Strength Uplift actually count?

It counts conversions Google can now attribute to your ads that it could not before. The purchase or lead already happened. Browser privacy rules, blocked scripts or a customer switching from phone to laptop had hidden it from Google.

• Enhanced conversions for web takes the email or phone number a customer types at checkout or into a form, hashes it with SHA-256 on your site, and sends the hash to Google. Google matches it against the hashes of Google accounts that were signed in when someone saw or clicked your ad.
• Enhanced conversions for leads does the same for leads that turn into customers later, when you upload them from your CRM or a spreadsheet.
• Google tag gateway serves Google's tag from your own domain instead of Google's, so measurement requests travel as first-party traffic, which browsers and blockers interfere with less.

None of these makes a single extra person buy anything. They change which of your existing customers Google can see and credit to an ad.

A recovered conversion is a sale that already happened. Google just found it.

03Are recovered conversions real sales?

Yes, and that is exactly the problem. They are real sales. What is not established is that the ad caused them.

The first gap is between what happened and what Google measured. Enhanced conversions narrows it. The second gap is between what Google credits to an ad and what the ad actually caused. Enhanced conversions does nothing for that one, and signed-in matching can widen it, because it credits ads for customers who were already on their way to you.

The two best studies of that second gap were run inside the platforms themselves. At eBay, Tom Blake, Chris Nosko and Steven Tadelis turned paid search off in controlled experiments. Brand-keyword ads showed no measurable short-term benefit. On non-brand keywords, ads did influence new and infrequent buyers, but most of the spend went on frequent buyers whose purchases the ads did not change, and average returns were negative. At Facebook, Brett Gordon and colleagues ran 15 randomised advertising experiments covering 500 million user-experiment observations and compared them with the standard non-experimental methods. In half of the studies, those methods got the sales lift wrong by a factor of three.

Sources: Google Ads Help, "About enhanced conversions"; Blake, Nosko and Tadelis, Econometrica, 2015; Gordon, Zettelmeyer, Bhargava and Chapsky, Marketing Science, 2019. Framework ours.

Neither study is recent, and neither tested enhanced conversions. They show how wide the gap between counted and caused can be, measured inside the platforms. Better counting does not shrink that gap.

At eBay, brand-keyword ads showed no measurable short-term benefit. Counting those conversions more accurately would not have changed that.

04Will enhanced conversions change what Smart Bidding spends?

Yes, and this is the part most advertisers miss.

Take a business spending ₹50,000 a month on Search with a target CPA of ₹500. Google reports 100 conversions. The order system shows 111, so Google is missing 11. Switch on enhanced conversions and assume it recovers all 11, in line with Google's 11% average. Google now reports 111, and your reported cost per conversion falls from ₹500 to ₹450. Nothing about the business has changed.

Now look at the target. It says ₹500, but you set it while Google was under-counting. Back then, a reported ₹500 meant a real ₹450. Smart Bidding now sees conversions coming in ₹50 under target, and it does what it is built to do: it bids higher and spends more until the reported number is back at ₹500. With counting fixed, a reported ₹500 now means a real ₹500.

Illustrative: ₹50,000 a month, 111 real orders, Google's 11% average applied in full. Your recovery rate and bidder behaviour will differ.

Whether ₹500 per real customer is acceptable depends on where the target came from. If you worked it back from margin, keep it: you are now bidding to the number you meant. If it was simply what the account delivered last quarter, it has just been loosened by the size of the recovery, and nobody decided that.

Google's own help pages say a change to conversion volume triggers a learning period of 1 to 2 conversion cycles, and advise setting budgets to the amount you actually intend to spend before you make the change. So set the budget first, switch on, re-set the target against your order data, and do not judge anything until two conversion cycles have passed.

Same target CPA, more counted conversions: Google gets permission to spend about 11% more on every real customer.

05Should a small business in India set up Google tag gateway?

Only if your site already sits behind a supported CDN. Google offers automatic setup through Cloudflare, and manual setup through other CDNs such as Akamai, Fastly and Amazon CloudFront, a Google Cloud load balancer, or a web server you control. Cloudflare says the feature itself is free.

Most Indian small-business sites run on none of those. Shared hosting on Hostinger, GoDaddy or Bluehost, a Shopify store or a Wix site gives you no CDN rules to edit and no server to configure. Our own site runs on Hostinger, and we do not use tag gateway either.

Moving your DNS to Cloudflare to get it is a real project: email records, SSL and every subdomain move with it, and one missed record means lost enquiries. That is too much risk for a 14% figure from a before-and-after comparison.

If your site is not already behind Cloudflare, tag gateway is a DNS migration sold as a checkbox.

06Is sending hashed customer data to Google allowed under India's DPDP Act?

Treat it as personal data, because it is. Hashing is not anonymisation. Enhanced conversions only works because Google can match the hash to a real, signed-in account.

Google's customer data policies already require you to tell customers their information is shared with third parties for ad measurement, and to get consent where the law requires it. Under India's Digital Personal Data Protection Rules, 2025, the obligations that cover your forms (notice, consent and security safeguards) apply from 13 May 2027. The 13 November 2026 date compliance vendors keep advertising only starts the registration of consent managers.

The practical step costs nothing: add a line to your forms and privacy notice saying you share contact details with advertising partners, including Google, to measure ad performance. Google requires it today and the law will by May 2027. This is not legal advice. If you handle health, finance or children's data, ask a lawyer first.

07What should you actually set up this week?

In this order:

1. Record a 30-day baseline. Orders or qualified leads from your own system, and conversions as Google Ads reports them. The gap is what you are trying to recover.
2. Turn on enhanced conversions for web. Use the Google tag's automatic option or a user-provided data variable in Google Tag Manager, and accept the customer data terms. About half an hour, and free.
3. If you sell through leads, import lead quality. Store the Google click ID or the form's email with each lead, then import your "qualified" and "won" stages through Data Manager. Google Sheets, HubSpot and Zoho CRM connect directly. Make the qualified stage your primary conversion. This is the one step here that changes what you pay for, not just what you see.
4. Set your budget, then re-set your target. Base the new target CPA or ROAS on your own order data, not on last quarter's reported number, then leave it alone for two conversion cycles.
5. Skip tag gateway unless you are already on a supported CDN.
6. Read the uplift metric as a measurement report. Never put it in a monthly report as growth. Judge the account on cost per real customer.

If you spend more than about ₹3 lakh a month, go further: switch ads off in two or three comparable cities for four weeks and compare sales with cities where they kept running. Meridian GeoX, the open-source library Google just made generally available, is built for this. It is the only method here that measures what the ads caused.

This is the same pattern as Google's AI Max migration earlier this month: the automation is only as good as the conversions you feed it. If you are not sure whether your tracking, your lead quality or your targets are the real bottleneck, that is what our free growth audit is for. See how the growth engine connects ads, tracking and CRM, or browse our services.