What enhanced conversions add to the conversion tag
Google documents enhanced conversions as a feature that can improve the accuracy of conversion measurement by supplementing existing conversion tags with hashed first-party conversion data sent from the website in a privacy-safe way. The key word is supplement. The conversion action, its trigger, counting rules and window are all unchanged and are set up exactly as the conversion tracking tutorial describes. What changes is the payload: alongside the event, the tag sends one or more pieces of customer data the person provided on the site, hashed before transmission.
Why that helps is a matter of how matching works. Ordinary tracking depends on the click identifier surviving from the ad click to the conversion: through redirects, through browser storage limits, through consent decisions and through the time between the click and the action. Each of those is a place the identifier can be lost. Where the person was signed in to Google when they interacted with the ad, the hashed email or phone number offers a second route to the same ad interaction. The conversion is then counted, and AttributionAssigning conversion credit to touchpoints. can assign it to the campaign that earned it rather than to nothing.
The effect on reporting is additive. Conversions the click identifier already matched are not counted twice; conversions that would have been dropped are recovered. The effect on bidding follows: Smart Bidding sees more of the conversions that actually happened, and its predictions of which auctions convert get better data. For a comparable server-side idea on another platform, the Conversions APIServer-side event sharing with Meta. entry describes Meta's approach; Google's version is tag-based for web and upload-based for leads. Whether to start with this feature or with a server container is the subject of the comparison of enhanced conversions and server-side tagging.
Enhanced conversions for web versus for leads
The two variants solve different problems and are configured separately.
| Row label | For web | For leads |
|---|---|---|
| The conversion happens | On the website, where the tag fires | Offline, after a form on the website |
| Data captured | Email, phone, name and address fields at the point of conversion | Hashed email or phone from the lead form at the time of the click |
| How Google matches | Hashed data plus the click identifier, at conversion time | Hashed data uploaded later, matched back to the original ad click |
| What it replaces or improves | Conversions lost when the identifier is missing | Offline conversion imports that depended on storing the click identifier in the CRM |
| Typical business | Retail, subscriptions, any on-site transaction | Lead generation with a sales process after the enquiry |
| Setup location | Conversion action settings plus tag configuration | Conversion action for imports plus a capture tag and an upload process |
For web is the simpler of the two. The person completes a purchase or sign-up, the page knows their email because they just typed it, and the tag sends a hashed version with the conversion. Nothing else in the measurement stack changes.
For leads addresses a long-standing pain of lead-generation accounts: the real conversion, a qualified lead or a signed contract, happens in a CRM days or weeks later. The documented older approach was to capture the click identifier at form submission, store it against the lead, and upload it with the outcome. That pattern is an offline conversion importUploading outcomes recorded outside the website, such as closed deals, to an ad platform. keyed by the click identifier. Enhanced conversions for leads instead captures hashed lead data at the moment of submission and lets you upload the outcome keyed by that same hashed data; Google matches it to the ad click on its side. The CRM no longer has to store a click identifier, which removes the most fragile link in the chain. A lifecycle marketing platform that already holds lead email addresses and stage changes is usually where the upload is generated from. Which stage to upload, and how it changes bidding in an enquiry-led account, is covered in the Google Ads lead generation setup.
How the data is hashed and matched
Google documents that the customer data is hashed using a standard SHA hashing function before it is sent, so Google receives hashed values rather than raw personal data, and that those values are matched against hashed data from Google accounts that were signed in when the person interacted with the ad. Two consequences follow for implementation.
First, hashing is deterministic, so the input must be normalised the same way on both sides or the hashes will never match. Google documents normalisation requirements for the fields it accepts: email addresses trimmed and lower-cased, phone numbers in an international format, names and address components treated according to its field rules. When the Google tag or Tag Manager does the hashing, normalisation is handled for you; when you hash yourself before an API upload, it is your responsibility, and a wrong format silently produces zero matches.
Second, the match depends on the person having been signed in to a Google account at the time of the ad interaction. Google does not publish match ratesShare of the records in an uploaded customer or account list that an ad platform matched to its own users or companies., and the share of a site's visitors who were signed in varies by market, device and audience. This is why the feature's effect differs between accounts, a point the later section returns to. Where no match is possible, the conversion simply falls back to whatever the click identifier could do; the feature never makes measurement worse.
What Google does with the matched conversion is the same as with any other: it is attributed under the action's model and window, counted according to its counting setting, and reported in the Conversions column if the action is primary. The comparison with GA4 figures is relevant here, because GA4 does not use this matching and the gap between the two systems often widens once enhanced conversions are live.
Setup options: Google tag, Tag Manager, API
Enhanced conversions for web is turned on in the conversion settings, at the account level or per conversion action, with a choice of how the data will be supplied. Google documents three routes.
Google tag. If the Google tag is placed directly in the page, the user-provided data can be collected automatically, by letting the tag detect fields on the conversion page, or manually, by specifying CSS selectors or JavaScript variables that hold each field. Automatic detection is quick to enable and fragile on complex pages; manual specification is reliable when the page structure is known.
Tag Manager (GTMTag management system.). The conversion tag is given a user-provided data variable that reads the fields from the page or the data layer. The cleanest implementation pushes the values into the data layer at the moment of conversion, so the variable does not depend on the page's markup. Because the Google Ads conversion tag in Tag Manager does the hashing itself, the raw values exist only in the browser. This is the route most sites should take and the one that survives redesigns best.
Google Ads API. For server-side implementations, the hashed data is sent through the conversion adjustment or upload endpoints. This suits businesses that already process conversions server-side, that need to send data the page never sees, or that are implementing the leads variant, where uploads come from a CRM or automation platform rather than the browser. Where the conversion data passes through a server-side Tag Manager container, the server-side tagging go-live checklist includes confirming that enhanced conversions data still arrives in Google Ads after the cutover.
For leads, setup is two parts: a tag that captures and hashes the form data when the lead is submitted, configured through any of the routes above, and an import conversion action that receives the later outcome. Uploads can be made manually, scheduled from a sheet, or sent through the API, and each upload carries the hashed identifier, the conversion time and the value. When the upload is manual, the offline conversion upload SOP turns it into a routine with a consent filter, a reading of the import results and a log of what was sent.
A mortgage broker closes the loop
A mortgage broker runs search campaigns to a quote form and has tracked form submissions as its only conversion for years. Roughly 1,400 enquiries a month arrive, of which about 90 become signed cases, and the campaigns have learned to buy enquiries from comparison-minded visitors who rarely proceed. The broker turns on enhanced conversions for leads through Tag Manager, pushing the hashed email into the data layer on submission, and creates an import action named "Signed case". Each week the CRM exports signed cases with the email and the signing date, and the file is uploaded. After six weeks the import action records around 70 matched cases a month, because not every applicant was signed in to Google at the click. The campaigns are switched to optimise towards signed cases with the enquiry action made secondary. Enquiry volume drops by about a fifth over the next quarter while signed cases stay level, and cost per signed case falls accordingly, because bidding now has the outcome that pays the bills.
Consent and privacy requirements
Enhanced conversions move personal data, hashed or not, from the site to Google, and the rules around that are the same rules that govern any processing of customer data. Google requires advertisers to accept its customer data terms before the feature can be enabled, and its policies require that you have the necessary rights and consent to share the data and that your privacy noticeThe published notice that tells people what personal data the business collects, why, and which tools and partners receive it; marketing keeps it current. discloses the sharing. These requirements are documented; the detail of what counts as adequate consent depends on the jurisdiction and is a legal question, not a tagging one.
In practice three things need to be true. The consent banner must gate the conversion tag so that denied consent prevents both the conversion event and the enhanced data from being sent; a tag that sends hashed data regardless of consent is a compliance failure, not a measurement win. The privacy notice must say that hashed customer data is shared with advertising platforms for measurement. And the people operating the account must understand that the raw values never need to be stored anywhere new: the hashing happens in the tag or before the upload, so the implementation should not create a fresh store of email addresses.
Where consent mode is in use, Google documents that conversions from users who declined can be modelled rather than observed. Enhanced conversions and modelling address different gaps: modelling estimates what could not be measured; enhanced conversions measure more of what can. They coexist, and the account's reported conversions after both are enabled include observed, enhanced-matched and modelled components.
Validating: diagnostics and what recording means
Google Ads shows a diagnostics tab on each conversion action. For enhanced conversions it reports whether the hashed data is being received, flags common implementation problems, and shows a status that moves through stages as data arrives and is validated. "Recording" means the feature is receiving usable data. It does not say how many additional conversions were matched, and Google does not break out enhanced-matched conversions as a separate column in standard reports.
Validation therefore has two parts. The technical check is in the diagnostics view and in Tag Manager preview mode: confirm the user-provided data variable is populated on the conversion step, that the fields are the ones you expect, and that nothing is sent on steps where no conversion occurred. Common failures are an empty email field because the data layer push happened after the tag fired, a phone number without a country code, and automatic collection picking up a search box rather than the checkout email.
The measurement check is a before-and-after comparisonA measure read before and after a change with no comparison group, so the season and everything else that moved in those weeks is credited to the change.. Note the conversion count and conversion rate (CVRShare of visits or clicks that convert.) for the affected actions over a stable period before enabling the feature, then compare after a settling period of similar length, controlling for anything else that changed. The conversion rate calculator makes the two periods easy to express consistently. A rise concentrated on campaigns and devices where the click identifier was weakest, such as long sales cycles or mobile browsers with strict storage limits, is the signature of the feature working. A uniform rise across everything suggests another change happened at the same time.
When it helps most and when it changes little
The feature recovers conversions that the click identifier was losing, so its value is proportional to what was being lost. Google does not publish typical improvements, and claims of a fixed uplift should be treated as marketing. The conditions under which the gain tends to be large are documented in principle and visible in practice:
- Long gaps between click and conversion, where the stored identifier expires or is cleared before the action.
- Browsers that restrict storage, which affects some mobile and privacy-focused browsers more than others.
- Consent regimes where many visitors decline, provided the person later consents at the point of conversion or the enhanced data is sent lawfully.
- Lead-generation businesses adopting the leads variant, where the gain is less about recovering lost conversions and more about measuring a conversion that was never tracked at all.
- Cross-device journeysA journey in which the ad interaction and the conversion happen on different devices, so that only a signed-in identity or modelling can connect them., where the click and the conversion happen on different devices and only a signed-in identity can connect them.
The gain is small when the site already matches most conversions: short sales cycles, mostly desktop traffic, a market with high consent rates, and no offline step. In those accounts the diagnostics will say recording, the counts will barely move, and that is the correct outcome rather than a failed setup.
Enhanced conversions also do not fix a wrong definition of the conversion, a trigger that fires on the wrong event or a primary action that measures the wrong thing. Those belong to the base setup covered by the measurement hub and the setup tutorial, and they should be right before this layer is added. The author's conversion tracking expertise page describes the order in which these layers are usually implemented on an account.
When the diagnostics flag a problem, or report recording while counts stay flat, the enhanced conversions troubleshooting page works through the causes from the diagnostics tab and a tagged test conversion.
Common mistakes
- Enabling the feature on a broken base tag. There is nothing to enhance.
- Hashing without normalising. Capitalised emails or local-format phone numbers never match.
- Collecting the wrong field automatically. A newsletter box or a search field instead of the checkout email.
- Firing the enhanced data before the data layer is populated. The variable is empty and the diagnostics look fine at a glance.
- Sending data regardless of consent. A compliance failure that also distorts measurement.
- Judging the feature by the diagnostics status alone. Recording is not the same as recovering.
- Expecting a fixed uplift. The gain depends on what was being lost, and in some accounts that is very little.
- Leaving the CRM storing click identifiers after moving to the leads variant. Two parallel upload routes double count.
