Data clean rooms, explained

A data clean room is a controlled environment where two organisations combine data for analysis, with technical restrictions preventing either from extracting the other's individual records.

What it is for

The classic case: an advertiser has customer data, a media owner or retailer has audience data. Both want to know the overlap, so the advertiser can measure whether advertising reached its customers, or reach people similar to them.

Exchanging the data directly would mean transferring personal data, with the legal and commercial risk that implies. The clean room allows the question to be answered without the transfer.

How it works, in outline

1. Both parties upload data in a hashed or encrypted form.
2. Matching occurs inside the environment, on identifiers neither party can read.
3. Queries are restricted to those the environment permits.
4. Outputs are aggregated, with minimum thresholds preventing identification of individuals.
5. Neither party can export the other's records.

Point four is the essential control: a query returning a count below a threshold, typically 50 or 100, is refused, because a small enough group is identifiable.

What it is used for

Retail media has driven much of the adoption: a brand wants to know whether its advertising on a retailer's platform reached its own customers, and the retailer will not hand over its shopper data.

The scale threshold

Matching requires volume. Below a few hundred thousand records the match rates are low, the minimum thresholds bite, and the analysis produces little usable insight.

Setup and operating costs are also substantial. For a mid-sized advertiser the honest assessment is that this is not yet a relevant tool.

The legal position

A clean room reduces the risk of unauthorised disclosure. It does not create a legal basis for the processing.

Both parties still need a lawful basis for using their own data in this way, the arrangement needs to be documented between them, and the purposes must be within what the individuals were told.

Treating a clean room as a route around consent is a misunderstanding that has already attracted regulatory attention.

The realistic assessment

For large advertisers working with large platforms and retailers, clean rooms are becoming the standard mechanism for measurement collaboration.

For everyone else, the useful takeaway is the principle rather than the technology: build your own first-party data, so that when this kind of collaboration becomes worthwhile you have something to bring to it.