While much of the conversation around data clean rooms has focused on enabling safe audience planning and targeting, the critical measurement area has remained largely overlooked.

Measurement involves some of the most sensitive customer data around (e.g., transactions, conversions, and media exposures) and that data is often shared across the open web without proper protection. In this blog, we review two types of measurement solutions, the Privacy-Enhancing Technologies (PETs) that support them, and why it’s high time to separate true data clean rooms from all the pretenders.

Data clean rooms in the crosshairs

Late last year, my boss Lauren Wetzel sounded the alarm against privacy washing from unscrupulous tech providers slapping the data clean room label on outdated technology or offering misleading 'privacy-first' solutions. A few weeks later, the FTC warned advertisers and publishers that most data clean room vendors today are less interested in protecting privacy than they are in facilitating the sale of customer data (ouch!), and that using a data clean room doesn't automatically shield them from "their obligations under the law or the promises they have made to consumers."

When AdExchanger's Allison Schiff reported on the FTC story at the time, she closed with an ‘in other news’ quip about using a wet toothbrush on a cat’s head to remind them of being licked clean by their mother. Coincidence or not, I can’t get that image out of my head now, so I’ve chosen to embrace it:

How can you tell if your data clean room provider is the real thing or just a wet toothbrush? By looking at how it does measurement.

You can’t manage what you don’t measure 

How many impressions did your campaign generate? How many clicks? What was your ultimate click-through rate, conversion rate, CPM, cost per acquisition, or revenue per click? Advertisers today need top-of-the-line measurement solutions (including audience verification, reach and frequency, attribution, media mix modeling and incremental lift) to assess the effectiveness of their marketing campaigns.

The IAB Tech Lab put it succinctly in its Guidance and Recommended Practices for Data Clean Rooms: Organizations today need measurement “to understand ROI and justify advertising spend.”

But in today’s fragmented advertising ecosystem, measurement often calls for media exposure data (typically in the form of individual impressions) from multiple data sources to be joined and matched with the advertiser’s conversion data (or its partners’ data if the advertiser has little first-party data to start with, like a CPG brand). And more times than not, one of the parties jumps the gun and shares sensitive personal data with the others. Or everyone gets lost on their way to the legal department, and nothing gets done.

Thankfully, a true data clean room can provide the matching capabilities, operational and contractual efficiencies necessary to fulfill all modern measurement needs without delay and without sharing, commingling, or transferring sensitive records.

The difference between in-platform and off-platform measurement

At InfoSum, we make a distinction between in-platform and off-platform measurement. It’s crucial to understand that distinction because they don’t necessarily require the same PETs.

In-platform measurement is when all the measurement calculations take place inside the data collaboration platform, and the end result is a static readout. Users may receive a scorecard as a PDF, in Tableau or some other data visualization tool, but what they’re receiving is essentially a report — not data. With off-platform measurement, the matching operations still take place inside the platform, but anonymized individual-level data is then exported to a different environment for further analysis. This allows users with data science resources to perform additional queries, run proprietary models on the data, or offer advanced measurement solutions to their clients without jeopardizing the identity of individual consumers.

How do we preserve privacy along the way? At InfoSum, we use PETs like pseudonymization to safeguard sensitive information; decentralized data processing and secure multi-party computation to eliminate data movement; differential privacy to add noise to query results; and privacy budgeting to limit the number of queries any one party can run in a 24-hour period and therefore prevent reidentification through brute force. And for off-platform measurement, we create a Private Path to outside partners by replacing all deterministic IDs at the data egress stage with anonymous, point-in-time, non-reversible synthetic IDs.

The best of both worlds

In-platform and off-platform measurement aren’t mutually exclusive solutions. Our platform is designed for non-technical users, allowing them to select matching keys, define an intersection criteria, and run preset queries in natural language for key measurement functions like reach and frequency, or incremental lift — all at the push of a button and without any knowledge of SQL. An advertiser or agency can use insights produced inside the platform to provide immediate feedback to a publisher and optimize a campaign in flight, and then hand detailed (de-identified) results off-platform to their data science teams for additional, post-campaign diagnostics.

Visionary CTV and retail media players are leading the charge by embedding best-in-class data clean room measurement solutions into their ad sales offerings. In the UK, ITV is working with Dunnhumby to provide quick and reliable performance insights to CPG brands advertising on ITVX and selling at Boots and Tesco. Similarly, Channel 4 has teamed up with Sainsbury’s Nectar360 to measure campaign results for brands like Pepsi and L’Oréal. And CTV platforms in the US are doing similar work with Polk Automotive, Catalina, and Circana.

Data collaboration has a bright future, but only if it’s based on rock-solid data clean room principles, not wet toothbrushes. I have a buyer’s checklist for you.