Since June 15, 2026, new offline conversion imports and enhanced conversions for leads uploads can no longer be built on the Google Ads API. They go through the Data Manager API instead, or through Data Manager inside Google Ads if you do not want to write code. Integrations that were already uploading through the Google Ads API before the cutoff can keep running while they migrate. For agencies, the practical rule is simple: every new client setup uses Data Manager or the Data Manager API, and every existing API-based setup gets a migration plan.
This guide covers what changed, who is affected, how the old fields map to the new ones, what the new API does differently (including one thing it does not support at all), and what to do across a client book. If you need the basics of offline imports first, start with our offline conversion tracking guide.
What changed, and when
Google announced the change on the Google Ads developer blog on May 15, 2026. The key points:
- From June 15, 2026, the Google Ads API stopped accepting new adopters of offline conversion imports, including enhanced conversions for leads.
- The Data Manager API is now Google's primary API for importing offline conversions.
- A new integration that calls
ConversionUploadService.UploadClickConversions gets the error CUSTOMER_NOT_ALLOWLISTED_FOR_THIS_FEATURE.
- Integrations that had already adopted offline imports or enhanced conversions for leads can keep using the Google Ads API while they move to the Data Manager API.
This follows the same pattern Google used earlier for Customer Match and for session attributes and IP address data, which also moved to the Data Manager API.
Who counts as "existing"
The allowlist is not tied to the advertiser account. Google's Google Ads API documentation says UploadClickConversions requests fail if your Google Cloud project had no offline conversion upload requests between December 17, 2025 and June 15, 2026. Older setups that still authenticate with a legacy developer token are checked by that token instead. Google Ads Help words the same rule as developer tokens that sent no requests between January and June 2026 not being allowlisted.
That has two consequences agencies miss:
- An internal upload script that was built but sat idle over that period is not allowlisted, even if it worked last year.
- Spinning up a new Google Cloud project for a new client, and pointing an old script at it, makes it a new adopter. It will fail.
Google's announcement does not give an end date for allowlisted integrations. Treat the allowlist as a migration window, not a permanent exemption.
What the Data Manager API is
The Data Manager API is a single Google API for sending first-party data to Google advertising products. For Google Ads conversions, you call one method, events.ingest (IngestEvents in the client libraries), with a list of destinations and a list of events. The same data model is used across Google products rather than one model per Google Ads use case.
Compared with the Google Ads API, Google's upgrade guide lists these differences:
| Area | Google Ads API | Data Manager API |
| Access | Developer token with an access level; test accounts only until approved | No access approval step; OAuth with the datamanager scope |
| Quotas | Based on access level | Based on the Google Cloud project |
| Errors | Partial failure: good rows go through, bad rows are reported | Fast fail: one invalid event fails the whole request |
| Adjustments | Separate ConversionAdjustmentUploadService with an adjustment type | Same IngestEvents call; a repeated transaction ID is treated as an adjustment |
| Retractions | Supported | Not supported |
| Encryption of user data | Not available | Optional, on top of hashing |
| Session attributes and IP address | Allowlisted users only | Available to all users |
The fast-fail model is the change that bites first. With the Google Ads API, a batch with one bad timestamp still uploaded the other rows. With the Data Manager API, it uploads nothing. Validate rows before you send them, and use validateOnly: true while building.
Data Manager in Google Ads: the no-code route
Not every client needs an API integration. Data Manager inside Google Ads connects data sources through point-and-click setup. Google's Data Manager help lists these sources for conversion imports:
- File upload (such as CSV) and Google Sheets
- SFTP, HTTP or HTTPS, Google Cloud Storage and Amazon S3
- BigQuery, Amazon Redshift, Snowflake, MySQL, PostgreSQL and Oracle
- Salesforce and HubSpot
For offline conversion imports, the required columns are the GCLID (required when the Google tag is not set up to collect user-provided data) and the conversion date and time. Conversion value, order ID, GBRAID, WBRAID, event source, IP address, user agent and session attributes are optional. If you upload a file, the first line must be the headers.
Details from Google's help that matter in practice:
- Lookback per run. File-type sources (Sheets, SFTP, HTTP, Cloud Storage, S3) import conversions from the last 90 days on every run. Salesforce and HubSpot import the last 14 days on the first successful run, then changes since the last run. Warehouse sources such as BigQuery import the last 14 days on every run.
- Time zones. Rows without a time zone need a fallback time zone in the connection settings. If none is set, those rows are not imported.
- Date order. An ambiguous date like 02/01/2026 may be read as month first. A day-first row can fail the import, so use ISO 8601 such as
2026-11-03T14:22:00-05:00.
- Fresh data. Data Manager does not refresh your source. A daily schedule needs the sheet or table updated before the run starts.
The older "Import conversions from ad clicks using files" flow in Google Ads is now labelled legacy in Google Ads Help, which points users to Data Manager instead. Existing scheduled file uploads are not part of the June 15 API change, but new client setups should start in Data Manager.
Field mapping: Google Ads API to Data Manager API
If you are migrating a script or middleware, this is the core of the work. The Data Manager API puts the account and conversion action on the destination, and the rest on the event.
Google Ads API (ClickConversion) | Data Manager API | Notes |
customer_id on the request | destinations[].operatingAccount with accountType: GOOGLE_ADS | Must be the account that owns the conversion action |
login-customer-id header | destinations[].loginAccount | The account your authenticating Google user has access through |
conversion_action resource name | destinations[].productDestinationId | The conversion action ID only; the action must be an "Import from clicks" (UPLOAD_CLICKS) action |
gclid, gbraid, wbraid | adIdentifiers.gclid, adIdentifiers.gbraid, adIdentifiers.wbraid | Grouped under ad identifiers |
conversion_date_time | eventTimestamp | RFC 3339, for example 2026-11-03T14:22:00-05:00 |
order_id | transactionId | Also drives deduplication and adjustments |
conversion_value, currency_code | conversionValue, currency | ISO 4217 currency code |
user_identifiers (hashed email, phone, address) | userData | Set encoding on the request to match how you hashed |
consent | consent on the request or the event | adUserData and adPersonalization; event-level overrides request-level |
custom_variables | customVariables | Unknown variables are ignored |
cart_data | cartData | Each item needs item ID, merchant product ID, unit price and quantity |
Two fields are new and easy to miss. eventSource is required for Google Ads offline conversions in the send-events guide (for a website lead, WEB). And the timestamp format changes: the Google Ads API wanted yyyy-mm-dd HH:mm:ss+|-HH:mm with a space, while the Data Manager API wants RFC 3339 with a T. A script that only swaps field names and keeps the old date string will fail every request.
A trimmed request body looks like this. IDs and values are placeholders.
POST https://datamanager.googleapis.com/v1/events:ingest
{
"destinations": [{
"operatingAccount": { "accountType": "GOOGLE_ADS", "accountId": "1234567890" },
"loginAccount": { "accountType": "GOOGLE_ADS", "accountId": "9876543210" },
"productDestinationId": "123456789"
}],
"encoding": "HEX",
"consent": { "adUserData": "CONSENT_GRANTED", "adPersonalization": "CONSENT_GRANTED" },
"events": [{
"adIdentifiers": { "gclid": "EAIaIQobChMI_example" },
"eventTimestamp": "2026-11-03T14:22:00-05:00",
"transactionId": "crm-deal-48213",
"eventSource": "WEB",
"conversionValue": 2400,
"currency": "USD"
}],
"validateOnly": true
}
Each event needs at least one identifier: an ad identifier such as gclid, gbraid or wbraid, session attributes, userData, or an IP address. A request can carry up to 2,000 events. After sending, check processing with requestStatus.retrieve and the diagnostics in Google Ads, because acceptance of the request does not mean every conversion matched a click.
Adjustments, and the missing retraction
In the Data Manager API, you do not send a separate adjustment. If you send an event with a transactionId that already exists for the same conversion action, Google treats it as an adjustment:
- Value restatement: a new
conversionValue replaces the recorded value. The conversion count does not change.
- User data enhancement: if the original conversion had no user data, the new user data is attached. If it already had user data, the new data is ignored.
- Other fields, such as the gclid, keep their original values.
The adjustment event still has to pass normal validation, so it needs an identifier like the original. Adjustments for store sales and call conversions are not supported.
Retractions are not supported. In the Google Ads API, a retraction set a conversion's value and count to zero, which lead-gen setups used when a closed deal was cancelled or a lead turned out to be spam. In the Data Manager API you can restate the value but not remove the conversion. If a client's setup depends on retractions today, you have three options: keep that part on the allowlisted Google Ads API integration while it lasts, change the process so conversions are only sent once a stage is reliable (for example, after a cooling-off period), or accept restating the value instead. Decide this before you migrate, not after.
Enhanced conversions for leads in 2026
The same June 15 change covers enhanced conversions for leads uploads. Google Ads Help also says enhanced conversions for web and for leads have been combined into a single account setting in 2026, and that Google Ads now accepts user-provided data from website tags, Data Manager and API connections at the same time.
Google recommends enhanced conversions for leads for new setups and as the upgrade path for existing offline imports. Its upgrade help suggests sending, where available, the GCLID, hashed email or phone, a unique order ID to prevent duplicates, and GBRAID or WBRAID. It also suggests running the new action alongside the old one and swapping which is primary after the longer of three conversion cycles or four weeks. Our enhanced conversions setup guide covers the website side, and the glossary entry has the short version.
What agencies should do now
New client setups
- Default to Data Manager in Google Ads. If the client is on HubSpot or Salesforce, or can get stage changes into a Sheet, warehouse table or SFTP file, the built-in connectors cover it without code.
- Use the Data Manager API only when the UI cannot. Custom backends, real-time sends, or a data volume or transformation that connectors do not handle.
- Create the API project in the right place. A Google Cloud project owned by the client or your agency, with the OAuth user given access to the Google Ads account (or the manager account above it).
- Plan for no retractions. Choose CRM stages that rarely reverse.
- Send a unique transaction ID on every conversion, so duplicates are absorbed and value restatements work later.
Existing client setups
- Inventory how every account imports today. Native CRM connector, Data Manager, scheduled file, third-party automation tool, or your own Google Ads API code. Only the last category, plus any third-party tool that calls the Google Ads API on your behalf, is directly affected.
- For third-party tools, ask the vendor whether they have moved to the Data Manager API and whether anything changes on your side. Do not assume.
- For your own code, confirm the Cloud project (or legacy developer token) was uploading during the allowlist window. Then build the Data Manager API path next to it and run both into separate conversion actions, or with
validateOnly, until counts agree.
- Check where conversion actions live. Many agencies use conversion actions owned by a manager account. In the Data Manager API,
operatingAccount must be the account that owns the conversion action. The Google Ads API also accepted a parent or child account; the Data Manager API does not.
- Rewrite error handling for fast fail. Pre-validate timestamps, identifiers and currency before sending, and split batches so one bad CRM record does not block a day's uploads.
Where agencies trip up
- Reusing an old script for a new client. A new Cloud project is a new adopter and gets
CUSTOMER_NOT_ALLOWLISTED_FOR_THIS_FEATURE.
- Keeping the old timestamp format. The space-separated Google Ads API format is rejected; use RFC 3339.
- Sending the conversion action name. The API wants the numeric conversion action ID in
productDestinationId, not the name a file upload uses.
- Using the wrong conversion action type. The action must be an import from clicks action. Sending to a website tag action by mistake is a different feature (tag conversions with an additional data source).
- Relying on retractions for spam or cancelled deals. They no longer exist on the new path.
- Day-first dates in Sheets. A UK or Australian client sheet with 03/11/2026 can fail or land on the wrong day.
When rows fail, our offline conversion import errors page lists the common causes, and the CRM import guide covers click ID capture and stage mapping, which have not changed.
Getting it done for a client
Most agencies only need to touch the accounts with custom API uploads, but those tend to be the biggest clients. If you'd rather have the migration built and tested for a client, send us one account and we will tell you what it uses today and what needs to move.