Identity and Privacy
Google Ads Data Hub and PAIR: how Google's clean room works
Published · Last updated
Next review:
Google Ads Data Hub is Google's clean room for advertisers and measurement partners. It joins your first-party data with event-level data from Google's ad platforms and returns only aggregated rows to your BigQuery project. Rows covering too few users are dropped; the usual floor is about 50 users (Google, ADH privacy checks).
Contents
At a glance
Company
Google LLC, a subsidiary of Alphabet Inc. (NASDAQ: GOOGL)
Headquarters
Mountain View, California, US (Alphabet business address, SEC EDGAR, accessed 9 November 2026)
Founded
Beta announced 24 May 2017 (PPC Land, AMC timeline)
Ownership
Public company product (Alphabet Inc.)
Product category
Walled-garden clean room for Google ad data
What it does
Runs SQL over Google campaign events joined with your data; writes aggregated results to your BigQuery datasets; builds audiences for Google Ads and Display & Video 360
Deployment model
Google-operated; requires your own Google Cloud project for inputs and outputs (Google, ADH introduction)
Pricing model
n/d (Google publishes no Ads Data Hub price)
Key integrations
Display & Video 360, Google Ads, Campaign Manager 360, YouTube, Search Ads 360, Google TV (since 8 July 2026) (Google, ADH release notes); measurement vendors such as Adform, Comscore, DoubleVerify and Dynata (Google, ADH vendors)
Publicly named customers
n/d (Google publishes no customer list)
Standards supported
PAIR, which Google Ads originated and IAB Tech Lab now maintains (IAB Tech Lab, PAIR)
Last verified
9 November 2026
| Field | Google Ads Data Hub |
|---|---|
Company | Google LLC, a subsidiary of Alphabet Inc. (NASDAQ: GOOGL) |
Headquarters | Mountain View, California, US (Alphabet business address, SEC EDGAR, accessed 9 November 2026) |
Founded | Beta announced 24 May 2017 (PPC Land, AMC timeline) |
Ownership | Public company product (Alphabet Inc.) |
Product category | Walled-garden clean room for Google ad data |
What it does | Runs SQL over Google campaign events joined with your data; writes aggregated results to your BigQuery datasets; builds audiences for Google Ads and Display & Video 360 |
Deployment model | Google-operated; requires your own Google Cloud project for inputs and outputs (Google, ADH introduction) |
Pricing model | n/d (Google publishes no Ads Data Hub price) |
Key integrations | Display & Video 360, Google Ads, Campaign Manager 360, YouTube, Search Ads 360, Google TV (since 8 July 2026) (Google, ADH release notes); measurement vendors such as Adform, Comscore, DoubleVerify and Dynata (Google, ADH vendors) |
Publicly named customers | n/d (Google publishes no customer list) |
Standards supported | PAIR, which Google Ads originated and IAB Tech Lab now maintains (IAB Tech Lab, PAIR) |
Last verified | 9 November 2026 |
What it does
Ads Data Hub answers questions that standard Google reports cannot. How many people saw a YouTube ad and then bought in your store? What is the reach across DV360 and Google Ads together? Which frequency produced the best conversion rate? You write the SQL, ADH runs it over Google's event data and your linked data, and the answer lands in your Google Cloud project (Google, ADH introduction).
Two groups use it. Marketers and their agencies measure their own campaigns and build audiences. Measurement partners run brand lift, viewability, brand safety and reach studies for clients; Google lists vendors such as Comscore, DoubleVerify and Dynata (Google, ADH vendors).
Google's own page says access runs through a Google account team or an agency, and describes Ads Data Hub as "in a closed beta" (Google, ADH for Marketers). It is therefore not a self-serve product for most advertisers.
PAIR is the second half of this profile. Publisher Advertiser Identity Reconciliation was "originally developed by Google Ads team" and is now an IAB Tech Lab standard. It lets an advertiser and a publisher match first-party IDs with commutative encryption, without either side decrypting the other's list (IAB Tech Lab, PAIR). Google first shipped it inside Display & Video 360; InfoSum announced a clean room integration with that version on 25 May 2023 (InfoSum, 25 May 2023). IAB Tech Lab has since retired the competing OPJA specification in favour of PAIR.
How it works
Google's raw data sits in a Google-owned Cloud project that you cannot see. Your data sits in your own project. ADH reads from both, applies privacy checks, and writes only aggregated output back to you (Google, ADH introduction).
The privacy checks are the strictest of the major clean rooms we profile (Google, privacy checks):
5 of 5 rows
Aggregation threshold | About 20 users per row with noise injection; about 50 with difference checks; about 10 for queries on clicks and conversions only |
Noise injection | Adds calibrated random noise to aggregates; Google's recommended mode |
Difference checks | Legacy mode that compares results with earlier jobs and filters rows that could single out users |
Data access budget | Limits how many times a given piece of data can be read |
Static checks | Examine query statements for obvious, immediate privacy concerns |
Aggregation threshold
- What it does
About 20 users per row with noise injection; about 50 with difference checks; about 10 for queries on clicks and conversions only
Noise injection
- What it does
Adds calibrated random noise to aggregates; Google's recommended mode
Difference checks
- What it does
Legacy mode that compares results with earlier jobs and filters rows that could single out users
Data access budget
- What it does
Limits how many times a given piece of data can be read
Static checks
- What it does
Examine query statements for obvious, immediate privacy concerns
Table by AdSovereign, CC BY 4.0.
Filtering is silent: a row can vanish "without warning", and some checks depend on your earlier results. To help, ADH includes a synthetic sandbox dataset that is not subject to the checks (Google, ADH for Marketers). These thresholds work like k-anonymity floors, while noise injection follows the logic of differential privacy.
Matching your data to Google's has several routes, each with limits (Google, join your data):
- Mobile device IDs: the cheapest set-up, for app-heavy advertisers.
- Custom Floodlight variables: limited to Google Marketing Platform IDs.
- Cookie matching: no longer offered for new onboarding.
- User-provided data matching: since 22 April 2025, via Private Cloud Match, with data staying inside your VPC Service Controls perimeter (Google, ADH release notes).
Audiences built in ADH go to linked Google Ads and DV360 accounts. A list needs 100 or more active users in the last 30 days to serve on network inventory and YouTube (Google, audience activation). ADH works through a web UI and an API.
Pricing
Google does not publish a price for Ads Data Hub, and the developer pages we read do not mention one. Because results are written to BigQuery datasets in your own Google Cloud project, storage and any further queries there run under your Google Cloud account. We could not confirm from Google's ADH pages whether ADH query processing itself is billed.
Strengths and limits
Strengths
- Event-level Google data. ADH is the only access point for Google ad events keyed on Google and device IDs, including YouTube (Google, ADH for Marketers).
- Strong default privacy. Thresholds, budgets and noise are enforced by Google, not configured by each partner.
- Output in your warehouse. Results land in BigQuery, next to your other data.
- Steady updates. Google added CTV device IDs (26 March 2025), SQL table creation (20 February 2026) and Google TV inventory (8 July 2026) (Google, ADH release notes).
Limits
- One-way. Google's data never leaves; ADH measures Google media, not your full plan.
- Silent filtering. Analysts must design around rows that disappear without notice, and Google warns match rates "are often lower than users expect".
- Access is gated. Entry runs through Google teams or agencies, and the product page still says "closed beta".
- Google Cloud required. You need a Google Cloud project, even if your warehouse is elsewhere.
Alternatives
- Amazon Marketing Cloud: the Amazon equivalent, with its own aggregation rules.
- Google DV360: where PAIR-matched audiences are bought.
- InfoSum: a clean room that integrated with Google's PAIR in 2023.
- AWS Clean Rooms and LiveRamp Clean Room: neutral rooms for data outside Google.
The PAIR flow is drawn step by step in data clean room architectures compared. Neutral platforms are compared on price and ML in AWS Clean Rooms vs Snowflake vs LiveRamp. More context sits in the identity and privacy hub.
What we could not verify
- Any Ads Data Hub fee, or whether ADH query processing is billed to the customer.
- Which Google products support the IAB Tech Lab version of PAIR today, rather than Google's original implementation.
- The current meaning of "closed beta" on Google's marketers site.
- Customer names; Google publishes none.
FAQ
How much does Google Ads Data Hub cost?
How many users must a row have in Ads Data Hub?
What is Google PAIR?
Sources
- Google, "Privacy checks", Ads Data Hub for Marketers (accessed 9 November 2026). https://developers.google.com/ads-data-hub/marketers/guides/privacy-checks
- Google, "Introduction", Ads Data Hub for Marketers (accessed 9 November 2026). https://developers.google.com/ads-data-hub/marketers/guides/intro
- Google, "Ads Data Hub for Marketers" (accessed 9 November 2026). https://developers.google.com/ads-data-hub/marketers
- Google, "Ads Data Hub release notes" (entries to 4 November 2026; accessed 9 November 2026). https://developers.google.com/ads-data-hub/release-notes
- Google, "Join your data", Ads Data Hub for Marketers (accessed 9 November 2026). https://developers.google.com/ads-data-hub/marketers/guides/join-your-data
- Google, "Build and manage audiences", Ads Data Hub for Marketers (accessed 9 November 2026). https://developers.google.com/ads-data-hub/marketers/guides/audience-activation
- Google, "Third-party vendors integrated with Ads Data Hub" (accessed 9 November 2026). https://developers.google.com/ads-data-hub/marketers/resources/adh-vendors
- IAB Tech Lab, "Publisher Advertiser Identity Reconciliation (PAIR)" (accessed 9 November 2026). https://iabtechlab.com/pair/
- InfoSum, "InfoSum announces Google PAIR integration" (25 May 2023). https://www.infosum.com/blog/infosum-announces-google-pair-integration
- PPC Land, "Amazon Marketing Cloud" topic page, with the Ads Data Hub beta date (accessed 9 November 2026). https://ppc.land/amc/
More sources (1)
- US Securities and Exchange Commission, EDGAR submissions record for Alphabet Inc. (accessed 9 November 2026). https://data.sec.gov/submissions/CIK0001652044.json
Changelog
- 9 November 2026: profile published.
Spotted an error? Tell us