From the AnthologyAI Blog

Do Clean Rooms Wash Away All Sins?

May 7, 2024

Sometimes, the names we give technologies reveal much more about our intentions and desires than they do about the ostensible features they provide. As an intellectual exercise, let’s zoom in on one of the buzziest such names in marketing today—the “Clean Room.”

Sometimes, the names we give technologies reveal much more about our intentions and desires than they do about the ostensible features they provide. As an intellectual exercise, let’s zoom in on one of the buzziest such names in marketing today—the “Clean Room.” 

Like much “commercial upcycling” of terminology created in academic and scientific contexts, the term [Data] Clean Room alludes to Willis Whitfield’s modern creation of a sterile engineered space that could filter out microscopic particles that threatened electrical and mechanical components, designed to mitigate the risk of “stray particles a few microns wide [that] could compromise the integrity of a circuit board of a nuclear weapon.” What an audacious scientific barometer to live up to. 

And so, in the social and regulatory context of the last decade, rife with unauthorized data intrusions, cyber-crime, and a growing concern about the privacy value exchange we make in return for the “free internet”, Willis’ 6x6x6 foot sterile room has become analogous with a “safe and neutral space for data collaboration and partnerships … without either party (or parties) having access to the other’s customer data.” Layers of other novel PETs (Privacy Enhancing Technologies) from encryption (the padlock) and differential privacy (the booby trap) to private matching (the doorman) have in truth created a powerful ecosystem for brands, retailers, technology companies and other media stakeholders to keep a semblance of business as usual in the most lucrative new market of the post-modern era. Most importantly, Clean Rooms mitigate the existential threat to the golden goose of data assets - first party data.

Simply put, first party data describes data generated (explicitly or otherwise - more on this later) from the relationship between a customer and a vendor of wares, services, or information. Since the effect of the EU’s GDPR framework went into place in 2018, followed closely by similar regulations across various US states, the UK, and China - two elements of “business as usual” came under heavy fire: third party data (passively or implicitly collected user data of “varying” provenance, particularly through browser cookies and mobile apps - marketed commercially for other purposes) and implicit disclosed consent (where tracking and data collection policies could be buried deep in the legalese of Ts and Cs and painted with broad brush strokes).

With the (often well-deserved) demise of the murkier parts of the data and advertising technology industries came a swath of collateral damage. 

  • Brands and advertisers now struggle to maintain even a fraction of the personalized customer communications, analytics and measurement that have helped them grow, and struggle to find customers and improve their experiences.
  • Publishers (many of them powered by the very journalists who keep democracy alive around the globe) face a precipitous decline in yield and revenue from being unable to adequately pass signal from their properties to ad buyers. 

Clean Rooms, both on paper and in reality, came to the rescue in the years that followed Google’s Ads Data Hub experiment in 2018. With a healthy assist from the lengthy palliative care phase of the third party cookie, the industry at scale was able to arrive at scale before all was lost. And hooray for that; otherwise, we had too much to lose as a society.

A quick glance at the agenda of any data-driven advertising industry event, a tech blog, or LinkedIn on a Wednesday afternoon confirms that Clean Rooms and the accompanying workflows are rapidly becoming the most hopeful topic and area of investment in the data industrial complex. But the question remains - they solve many problems, but do they single-handedly rectify the fears that drove all this in the first place?

The answer is nuanced and not entirely ours to answer here at Caden. But let’s finish up by examining a couple of rhetorical gaps that need to be addressed before we “move on to the next issue” from privacy with our trophies in hand:

  1. Provenance and Consent: We need to decide what consent really means. Meta’s 2023 rang in with a bang and a €410mm fine for the fallacy of “forced consent.” There is growing language of “value exchanges” regarding consent, but what makes that value exchange a viable market is likely something more than expecting gratitude for personalization. Delta Airlines’ more magnanimous approach in their billion-dollar free wi-fi investment seems likely much closer to something scalable, and of course, the approach we’ve leaned all the way in on here at Caden. As an upstream (and downstream) factor, this isn’t a missing piece of the clean room model but the only one that helps deliver on the very promise of the word “clean.”
  2. Scalability: the first party data renaissance has had plenty of challenges that require creative solutions to the scale problem of marketing - the most obvious of which is the need to figure out how to scale audiences for brand and acquisition marketing without a huge war chest of cookies and signal to model against. I’m not addressing that here - the more interesting challenge is finding procedural efficiency in executing data collaboration without making every engagement a science project and a business development quagmire. Snowflake, AWS, Infosum, and Liveramp have done a lot to break down barriers and create seamless client experiences - but they have their own business and business models that intersect with each other, meaning they can never solve those challenges alone. Democratizing collaboration will likely take an outsider (or two!), the same way consent will, as we’re pretty convinced of here. Drawbridge founder Kamakshi Sivaramakrishnan is blazing a path with her new collaboration startup Samooha that works across clouds and providers - and she won’t be the last. 

In the end, we have much hope here at Caden that nominative determinism will treat Clean Rooms the way it did Usain Bolt. However, it won’t be altogether organic; it will take more than the capacity of the cloud and identity players to deliver on the promise of a post-cookie nirvana that balances scale and signal with the agency and rights of the consumer. From our perspective, the answer isn’t that mysterious. It will only come with bringing the Mountain to the Consumer, with transparent value exchanges and explicit consent. It’s (probably) what Willy would have wanted.

Partner with AnthologyAI

We work with the most progressive, technology-enabled, ethically-minded corporations who want to rethink how we collect and make decisions from consumer behavioral data. Our platform fulfills hundreds of different use cases. Want to see how we can bring new consumer intelligence to your business? Get in touch with our strategic development team.