Version 2026.43
3 breaking15 new7 changed3 retiring30 fixed
Google Ads API v25 migration¶
Google Ads, Google Ads Keyword Planner, Google Ads Audiences and Local Services Ads now use Google Ads API v25. Lead contact details no longer include email, and some Keyword Planner Forecast settings and columns are removed. Reselect “Keywords”, renamed from “Biddable keywords”, in saved configurations. For more information, see Google Ads: Setup guide.
TikTok Ads Reservation Ads and catalog dimensions removed¶
TikTok Ads no longer supports the Reservation Ads service type; basic report datastreams using it cannot be saved and fail to fetch. The catalog_id, product_id, product_set_id and sku_id dimensions are removed from Dynamic Product Ads (Catalog) reports. Update affected datastreams. For more information, see TikTok Ads: Setup guide.
Redirects and timeouts for notification deliveries¶
Slack, Microsoft Teams and webhook deliveries no longer follow redirects, and time out after 30 seconds. Update any target URL that answers with a redirect to its final address, and make sure receivers respond within 30 seconds. For more information, see Managing user notifications.
HubSpot Email Events additional columns¶
The HubSpot Email Events and Campaign Statistics report now includes the columns id, portalId, appId, emailCampaignGroupId, filteredEvent, sentBy.id, sentBy.created and url. The url column is filled only for CLICK events. For more information, see HubSpot: Setup guide.
LinkedIn Company Intelligence connector¶
The new LinkedIn Company Intelligence connector fetches company-level paid and organic engagement data by ad account, with an optional Campaign filter.
Twitter Ads MRC video metrics¶
Twitter Ads now offers six MRC-standard video metrics, including video views at least 50% in view for 2 seconds. Datastreams that already collect video metrics have them selected automatically, so their extracts gain the new columns. For more information, see Twitter Ads: Setup guide.
Datastream tag tools in MCP¶
The MCP server has new tools to search, assign and delete datastream tags. Tag changes appear in the datastream History tab. For more information, see Using Adverity MCP.
Tag and job state filters in MCP lists¶
MCP datastream lists can be filtered by connector type, and failed-job lookups by job state. List actions with a workspace now include child workspaces; set include_child_workspaces to false to limit results. Extract and authorization lists for a connection type now respect the workspace filter. For more information, see Using Adverity MCP.
Administration button for Viewer and Dashboard Manager roles¶
Users with only the Viewer or Dashboard Manager role now see an Administration button that lists the Administration pages they can open. Users without Administration access are taken to Overview instead of a Not found page. For more information, see Managing user permissions.
Failing since badge for notification subscriptions¶
Administration → Notifications now shows a Failing since badge for Slack, Microsoft Teams and webhook subscriptions whose delivery is failing, with the error type in a tooltip. The badge clears after the next successful delivery. For more information, see Managing user notifications.
Insert button for Custom Script references¶
Custom Script fields that take a column, another datastream, a value table or a workspace mapping now have an insert button that opens a searchable list. The saved instruction is the same as a typed one. For more information, see Using custom script transformations.
Reserved _divider instruction in custom scripts¶
In custom scripts, the instruction name _divider is reserved as a separator between instructions. It is not executed and is kept when transformations are imported. For more information, see Using custom script transformations.
Timestamp data type for fields¶
Fields can now be set to the data type Timestamp, which delivers timezone-aware timestamps natively to BigQuery, Snowflake, Postgres, Redshift, Databricks and SQL Server. Values are normalised to UTC, and time zone abbreviations such as EST are rejected; use a UTC offset. Existing fields are unchanged. For more information, see Create or Edit Columns.
Use key columns mode in the duplication monitor¶
The duplication monitor has a new Use key columns mode that compares rows on the key columns set in the datastream’s data mapping. Later changes to those key columns are applied automatically, so no separate column list is needed.
Warehouse deletion history¶
Warehouse now keeps a deletion history that logs every data deletion attempt. Each entry shows who started it, which data was targeted, and whether it completed, failed or was denied. For more information, see Introduction to the Data Warehouse page.
Datastream tags in the API¶
The API supports custom datastream tags, each owned by one workspace. New endpoints create, list, rename and delete tags, and bulk add, remove and read tags on up to 500 datastreams. The datastream list returns tags and accepts tag and tag_match filters. For more information, see Listing datastreams in Management API.
New endpoints:
GET, POST
/api/v1/datastream-tags/DELETE, GET, PATCH
/api/v1/datastream-tags/{tag_id}/DELETE, POST
/api/v1/datastreams/tags/POST
/api/v1/datastreams/tags/query/
Permission presets for API keys¶
API key permissions can be set to Read and write, Read-only or Custom, each limited to what you can do in the chosen workspace. The Create API key dialog and MCP sign-in now start on Read-only; choose Read and write if you need write access. For more information, see Authorizing Management API.
Timestamp data type in the API¶
Column and stack schema responses can now return the data type Timestamp, returned as data_type 13, and requests can set it. Values use the format YYYY-MM-DDTHH:MM:SS±HH:MM and are converted to UTC. Existing columns are unchanged. For more information, see Configuring Data Mapping in Management API.
Adform (Seller) report page reuse on retry¶
Adform (Seller) report pages are now stored in shared storage, so a retried fetch can reuse pages collected before the retry.
LinkedIn Lead Gen Forms API version 202609¶
LinkedIn Lead Gen Forms (Lead Sync) requests now use LinkedIn API version 202609 instead of 202602. No migration is needed, and existing datastreams keep their configuration. For more information, see LinkedIn Lead Gen Forms (Lead Sync): Setup guide.
Yahoo! JAPAN Display Ads API v19¶
The Yahoo! JAPAN Display Ads connector now uses Display Ads API v19 in place of v18, which is reaching end of support. No datastream reconfiguration is needed.
Authorization step in setup wizards¶
In the datastream and destination setup wizards, the Authorization step now hands over to a new authorization wizard when a new authorization is needed. Leaving a setup returns to the Select datastream type or Select destination type step with a notification.
Notification email design¶
Notification emails now follow the recipient’s light or dark mail setting where the mail client supports it, and use the current Adverity branding, including logo, colours, typography and buttons.
Scheduling terminology¶
Scheduling wording is aligned across the schedule box, scheduling dialogs, job trigger badges and the datastream list. The switch reads Scheduling, Bundle scheduling is called Smart Bundle, and Custom Schedule, Smart Schedule and Smart Bundle are capitalized as named modes.
Workspace filter no longer selects sub-workspaces¶
In the Workspace filter of the Datastreams, Authorizations, Destinations and Transformations tables, selecting a workspace no longer selects its sub-workspaces. Each selected workspace gets its own filter chip, and the table shows only the workspaces selected.
Criteo Retail Media legacy report types¶
Criteo Retail Media adds Performance and Attributed Transactions report types on API version 2026.07. Seven older report types are labelled Legacy and are being phased out; move to the newer report types. For more information, see Criteo Retail Media: Setup guide.
ironSource Partner connector deprecated¶
The ironSource Partner connector is deprecated.
WeChat connector deprecated¶
The WeChat connector is deprecated and is being phased out. For more information, see WeChat: Setup guide.
Adform (Buyer) settings page with legacy authorization¶
The Adform (Buyer) datastream settings page loads again for datastreams that use legacy username and password authorization.
Adform (Seller) dimension and metric lists¶
Adform (Seller) datastream settings load dimension and metric lists again. If a message about insufficient permissions appears, re-authorize the Adform (Seller) authorization.
Adobe Analytics 2.0 long metric names¶
Adobe Analytics 2.0 metadata updates no longer fail when a metric or dimension name is longer than 500 characters, so report-suite metadata refreshes complete.
Amazon Selling Partner weekly Brand Analytics fetches¶
Amazon Selling Partner Brand Analytics weekly fetches now split date ranges into Sunday-to-Saturday slices, so multi-week fetches no longer fail for all weeks except the last.
ChatGPT Ads retries on temporary 403 responses¶
ChatGPT Ads fetches now retry when the API returns a temporary 403 server error instead of failing immediately as unauthorized.
Criteo Marketing Solutions retries on server errors¶
Criteo Marketing Solutions fetches now keep retrying temporary Criteo server errors such as 504 Gateway Timeout for longer, with increasing waits, instead of failing after about a minute.
Facebook Ads breakdown settings form¶
Selecting the deprecated action_reaction action breakdown or removing the last breakdown in the Facebook Ads settings form no longer causes a script error, and incompatible breakdown options are disabled again.
Facebook Ads datastreams with Degrees of Freedom ads¶
Facebook Ads datastreams no longer fail entirely when an ad uses Degrees of Freedom and its Asset Feed Spec does not match its Object Story Spec; they return data from the other API calls.
Google Display & Video 360 reservation line item metadata¶
Google Display & Video 360 reservation line items that the SDF skips now return Line Item metadata such as Budget Type, Budget Amount, dates and pacing instead of blank values.
Google Drive files with duplicate names¶
Google Drive Process all now collects every file that shares a name within the same folder and creates one extract per file.
LinkedIn Ads fetches with endless pagination¶
LinkedIn Ads fetches no longer hang until timeout when LinkedIn keeps returning further pages; they now stop with a clear error after a very large number of pages or a repeated page token.
Marmind fetches with key figures or custom fields¶
Marmind fetches and metadata refreshes no longer fail when key figures or custom fields are selected.
Media Screenshots status checks¶
Media Screenshots status checks no longer fail with retry errors and run faster.
PubMatic access token refresh¶
PubMatic access tokens are now stored and refreshed before they expire instead of generated on every fetch, so accounts no longer approach PubMatic’s token generation limit.
Similarweb v5 report columns¶
Similarweb v5 reports now include the request’s domain, country, granularity and other settings as columns, plus base_domain holding the entered domain, as v4 reports did.
Sklik rate limit handling¶
Sklik data fetches no longer fail on HTTP 429 Too Many Requests responses; the connector now retries those requests automatically.
TikTok Ads location sync timeouts¶
TikTok Ads location sync no longer times out on authorizations with many advertisers.
TikTok Ads retries on temporary timeouts¶
TikTok Ads now retries temporary TikTok-side timeouts instead of reporting them as missing advertiser permissions.
Browser extension login through Adverity SSO¶
Browser extension logins through Adverity SSO complete again instead of stopping at the Sign in with Adverity SSO page.
Custom Script instructions page loading¶
The Custom Script instructions page no longer crashes when the instruction schema has not loaded; it shows the loading indicator and the instructions appear once loaded.
Custom script skipuntil with a field name¶
A skipuntil instruction with a field name now looks for a matching row below the header before falling back to the header cell; files without a preamble keep all rows.
Datalake files removed when pruning extracts¶
Pruning deleted extracts and clearing fetch artifacts now removes their files from the workspace datalake instead of leaving them behind.
Destinations outside the datastream workspace hierarchy¶
Destinations outside a datastream’s workspace hierarchy can no longer be added to it, and opening an existing binding of this kind now explains the problem instead of showing a generic error.
False unused subtable warnings in transformations¶
Transformations using mailto, collect, set_meta, set_tags or dump no longer show false warnings that subtables are unused.
Fetch card after saving a bundle schedule¶
After a bundle schedule is saved in the Set custom schedule dialog, the Fetch card on a fork datastream’s overview now shows the saved schedule as soon as the dialog closes.
Mass reload projects completing¶
Running mass reload projects no longer get stuck and now complete.
Pages failing to load after a new version¶
Pages that fail to load after a new version is deployed now reload once to get the latest version; if loading still fails, the actual error is shown.
Single Sign-On identity provider rows¶
Clicking a Single Sign-On identity provider row in Administration now opens its edit drawer for users who can manage SSO.
Support widget font loading and ticket details¶
The support widget no longer leaves text invisible while its font loads, and support tickets again include the product module, company ID and source page.
Workspace switch on the Data Dictionary page¶
Switching workspace from the Data Dictionary page now opens the Data Dictionary of the selected workspace instead of reloading the old one.