> For the complete documentation index, see [llms.txt](https://docs.mediarithmics.io/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.mediarithmics.io/collaboration/plan-and-integrate/setup-clean-rooms/security-and-privacy.md). # Security and privacy The privacy-enhancing techniques behind every Data Clean Room matching result and insight. ## Your goal Learn how the Data Clean Room module protects your data and your partner's, before you decide to share datasets or commit to a partnership. ## Differential privacy The Data Clean Room module uses **privacy-enhancing techniques (PETs)** to protect sensitive data while enabling secure collaboration. **Differential privacy** is a robust framework designed to provide strong privacy guarantees for individuals within a dataset. When performing aggregated transformations on two similar datasets that differ by only one record, differential privacy guarantees that the results will be nearly identical - added noise makes it difficult to determine whether any single individual's data is included in the dataset.

Matching and Insights on a dataset with a given user produce approximately the same result as on a dataset without that user — Adding or removing one user does not meaningfully change the output

This matters specifically for Clean Rooms because they are secure environments where multiple parties collaboratively analyze data without exposing their raw data to each other; differential privacy is what keeps the aggregated output from leaking anything about a specific individual. ### Noise introduction One of the key mechanisms of differential privacy is the introduction of noise. This noise is drawn from a probability distribution (a Laplace distribution, in mediarithmics' case) and added to the data or the query results. The noise obscures the presence or absence of any single individual's data in the dataset, so that no entity can infer sensitive information about individuals from the combined datasets. ### Privacy budget The **privacy budget** quantifies the total amount of privacy loss from queries on a dataset - in the Data Clean Room module, the queries required to calculate a number of matched records or to generate insights. Each query or analysis consumes a portion of this budget, with more sensitive queries consuming more. Careful management of the privacy budget keeps the cumulative risk of re-identification within acceptable bounds, even as multiple analyses are performed on the same data. ### Minimum threshold A **minimum threshold** dictates the minimum number of individuals that must be represented in any aggregate data release, to prevent releasing results specific enough to expose individual data points. If the number of users meeting a specific criterion is below this threshold, the result is suppressed or aggregated further. The minimum threshold is set to **10** by default: if fewer than 10 users are matched between your dataset and your partner's, a 0% match rate is returned instead. ### Rounding **Rounding** query results to the nearest specified increment makes it harder to infer precise information from the noisy output, and standardizes results across queries. Rounding is currently set to the **lowest ten** - if the actual number of matched records is between 55,340 and 55,349, the value returned is 55,340. ## What this means when you use a Clean Room * Aggregated information displayed as a result of a datasets matching - match rate, matched-record counts, Insights dashboards - is differentially private, not raw. * Neither partner can query or view individual record data imported into a Clean Room directly. * Only a small sample of the data you import is visible to users within *your own* organisation, for context on the dataset you created. ## Related pages


Analyze cross-partner behavior	Where match rate and Insights, both differentially private, are produced.	Analyze cross-partner behavior
Clean Rooms	The Collaborative App, Clean Room Template, and Clean Room layers.	Clean Rooms

--- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.mediarithmics.io/collaboration/plan-and-integrate/setup-clean-rooms/security-and-privacy.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.