Google Ads tightens tracking rules for Conversion Lift studies
Google has rewritten its help page for setting up Conversion Lift based on users, and the new version asks a good deal more of an advertiser's conversion tracking before an incrementality study can produce reliable results. The page now details the hashed customer data a tag must send, lists GCLID-only offline imports and GA4-imported web conversions as setups that do not qualify, and recommends 7 to 14 days of continuous conversion data plus roughly 75 to 80% of web conversion traffic routed through Google tag gateway before launch.
The change surfaced on 13 November. Google's page shows no revision date, so the exact day of the edit is unverified. The study mechanics are the same as before; the rewrite adds requirements and troubleshooting tables for the data that feeds them.
Conversion Lift is still not open to every account. Google says advertisers need to contact their Google account representative to use it.
Test reported conversions with a Conversion Lift study
A Conversion Lift study is a controlled experiment. Google splits an audience into a treatment group that sees the ads and a control group that does not, then compares conversions between the two. The difference is the lift, meaning the conversions the ads caused.
Standard reporting credits conversions to ad interactions whether or not the ad caused them, so it can run above the causal number. Meta Ads view-through inflation reached a median of 2.74x on purchases in consumer goods accounts, which is the kind of gap a lift study exists to expose.
Google offers two versions. The user-based study builds its groups from aggregated user attributes, so results can be segmented by factors such as age and gender. The geography-based study builds them from regions and supports offline data. For user-based studies, Google reports incremental conversions, relative lift, incremental conversion value, incremental cost per action and incremental return on ad spend.
Do GCLID imports work for Conversion Lift?
The biggest practical change is for lead-generation advertisers who upload offline conversions from a CRM. The updated page says imports matched only on the Google click ID are insufficient for a user-based study. Google wants enhanced conversions for leads instead, which send hashed first-party customer data such as an email address or phone number with the conversion.
GA4-imported web conversions appear on the list of unsupported setups. An account whose main conversion action is a GA4 key event imported into Google Ads will need a native Google Ads conversion action for the study.
GA4 imports have their own capture problems in any case. One audit found GA4 missed Christmas orders for a retailer and recorded only 32% of them after its consent banner switched itself off.
App conversions get their own detail, with requirements for Android, iOS, Firebase and third-party app attribution integrations. App campaigns targeting iOS and Travel Ads remain unsupported.
Advertisers face the same Conversion Lift budget and volume minimums
Google's existing recommendations on budgets, conversion volume, holdout sizes and campaign selection are unchanged. An account needs at least 1,000 observed conversions, and conversions with supplementary data do not count toward that figure. The study requires a minimum campaign budget of US$5,000 and at least one compatible conversion action. Reporting notes on supplementary conversions and the Demand Gen cooldown were already on the old page.
Display, Search, Video, Demand Gen, App and Performance Max campaigns are eligible. Each campaign can be in only one lift study at once, so a campaign already in a Brand Lift or Search Lift study has to come out of it before joining a Conversion Lift study.
Before launch, Google's feasibility tool shows a certainty estimate from 50% to 95% in 5% steps. Google advises aiming for 90% or more and treats anything between 50% and 90% as directional. If the setup falls below 90%, the tool suggests a budget that would get there.
Studies can run as short as 7 days, but Google recommends more than 14, especially for businesses with a long gap between impression and conversion. A study with a long conversion lag that ends inside 14 days can lose up to 17% of its measured absolute lift. Holdouts can go as high as 50%, and Google says most advertisers run one or two studies a year.
New diagnostic tables flag enhanced conversions errors
The longest new section is a set of diagnostic tables. Some checks appear in Google Ads under Goals, then Conversions, Summary and Diagnostics. Others are custom Conversion Lift checks that do not surface there.
The flags Google lists include:
- Enhanced conversions turned off for an eligible web action, or turned on without the account-level customer data terms signed.
- Web and app conversions arriving under the same conversion action. Conversion Lift drops all data for that action when it detects multiple sources.
- Tag pings missing user-provided data, or missing the email parameter on a large share of fires.
- Formatting errors. Emails must be lowercased, trimmed and SHA-256 hashed, phone numbers must be in E.164 format with the country code before hashing, and addresses need first name, last name, postcode and country.
- Relying only on automatic detection of user-provided data. Google supports it but says manual in-page code or an API delivers more.
Server-to-server uploads through the Google Ads API have a parallel list. The tag and the API must send the same transaction ID, uploads must land within 24 hours of the conversion, and OAuth credentials must be valid. Another check flags API uploads where hashed customer data is on fewer than 30% of pings relative to tag conversions (Google recommends sending it on every ping).
The page also recommends passing dynamic conversion values with valid currencies. That matters for incremental conversion value and incremental ROAS, which Google only reports when values are assigned to the conversion actions.
Campaigns targeting only Australia can skip advanced consent mode
Advanced consent mode sits on the readiness list, but Google scopes it to campaigns targeting the European Economic Area. The expanded guidance covers EEA, UK and Digital Markets Act scenarios. An Australian advertiser running Australian-only campaigns is outside that scope; one with UK or EU campaigns in the same account is inside it.
Google tag gateway is the other readiness item. Google describes it as a server-side setup that improves data collection and gives more control over advertising tags, and the updated page recommends routing about 75 to 80% of web conversion traffic through it before a study.
Two best practices carry over from the old version. Campaigns on Target CPA, Maximise Conversions or Target ROAS show higher lift, as do campaigns using data-driven attribution. Google's rewritten Smart Bidding help page caps Smart Bidding Learning status at 7 days, so moving a campaign to conversion-based bidding a few weeks before a study leaves time for it to settle.
Geo studies accept offline conversions by postcode
An advertiser whose conversions live offline, or who cannot get hashed customer data flowing, has another route. Conversion Lift based on geography accepts offline conversions aggregated to postcode or city level, alongside the Google Ads conversion tag, Firebase and Floodlight. Google says it can also measure multiple channels and multiple conversion types in one study.
The geo version has its own constraints. Campaigns must target a single country, the split uses Google Marketing Areas chosen to limit people converting in a different region from the one where they saw the ad, and only holdback and go-dark studies are available in the beta. Feasibility comes back as High, Medium or Low, and Google advises against proceeding on Low. The single-country rule is no obstacle for a business that only advertises in Australia.
Clear the Diagnostics flags before requesting a lift study
Because Google limits access to accounts that go through a representative, the first conversation will go better if the account already passes the checks. In order:
- Open Diagnostics under Goals and Conversions and clear every enhanced conversions flag on the web actions you plan to measure.
- Confirm customer data terms are accepted at account level.
- Replace any GCLID-only offline import with enhanced conversions for leads, hashed to Google's formatting rules.
- Make a native Google Ads conversion action the primary action if the account currently leans on a GA4 import.
- Split app and web conversions into separate actions.
- Pass real values and a currency on each conversion.
- Let 7 to 14 days of clean, continuous data accumulate, then ask about feasibility.
Enhanced conversions also recover unobserved conversions, so the audit is likely to help everyday reporting even if a study never runs. Start with the Diagnostics tab this week and work down the flags before anyone books a lift test.




