By 
Did you ever wish you could talk to your clean room?
Samooha is a data clean room provider startup co-founded and helmed by Drawbridge founder and former CEO Kamakshi Sivaramakrishnan. Last month, the company, which is backed by Snowflake Ventures, launched a conversational data collaboration widget on Slack and Microsoft Teams.
Brands, publishers and their partners can use it to converse in natural language, ChatGPT-style, with Samooha to answer questions, such as “What are my top campaigns based on impressions?“ or “Can you filter my recent campaign results by female users?”
The purpose is to make it easier for marketers who don’t have a data science background to securely access and share data regardless of where it sits.
Because the dirty secret about clean rooms is they can be rather difficult for nontechnical people to use.
Angling for easy
Although cloud providers like Google, Amazon and Microsoft bake privacy-enhancing technologies (PETs) into their solutions, businesses still require some level of data science expertise to take advantage, said Sivaramakrishnan, who spent more than two years in product at Microsoft-owned LinkedIn after the Drawbridge acquisition in 2019.
“A lot of what you might call misbehavior with data isn’t intentional,” she said. “It’s happening, at least partly because there isn’t an easy button for enterprises to do right by their customer data.”
Samooha – which is a play on the word for “group” or “community” in Sanskrit, if you were wondering – has two product experiences.
There’s a “developer edition” that gives businesses the option to upload their own data models and run advanced analytics and a no-code option with templates for secure data collaboration across different clouds, including matching and activation.
Starting this week, Samooha added a new freemium tier to help companies play around with both versions of the data collaboration experience before upgrading to a paid account.
“The idea here is to bring a real SaaS [software-as-a-service] dynamic to clean rooms,” Sivaramakrishnan said.
Data escrow
But for a SaaS-based model to work, there needs to be interoperability between different cloud providers.
Samooha uses a cryptographic technique called multiparty compute that allows for pooling multiple data sets and runs joint computations while keeping each party’s underlying data separate and private from the other.
“The data never has to move anywhere,” Sivaramakrishnan said. “We bring the compute to the data.”
Data collaboration without the need for data migration is something she had been thinking about since her cross-device days at Drawbridge in the mid-2010s. Years before the data clean room market started to take off, clients were already asking for clean rooms – they just weren’t calling them “clean rooms.”
“We’d get questions all the time, like, ‘Hey, is there an escrow-like environment we can use so we can share data without actually sharing data?’” Sivaramakrishnan said.
Drawbridge ended up developing a so-called “private graph” solution so marketers could combine their first-party data with Drawbridge’s cross-device data in a locally hosted graph that only they could access.
Although it was secure, scale was a challenge, which is why it makes sense for privacy-enhancing technologies, like multiparty compute, to be built natively into cloud solutions and for clean room apps to be able to run agnostically on any platform, Sivaramakrishnan said.
“PETs can’t come as an afterthought,” she added.
No excuses
And, increasingly, they aren’t.
PETs are part of the conversation, as the online advertising industry attempts to thread a very tricky needle, which involves responding to an uptick in regulatory scrutiny and rolling with platform privacy changes – while still somehow managing to do business.
The IAB Tech Lab launched a PETs working group last year to develop standards, which is part of the Tech Lab’s ongoing Project Rearc initiative.
Though many of Samooha’s clients are advertising and marketing companies, Sivaramakrishnan and her co-founder and CTO, Abhishek Bhowmick, have ambitions of tapping into sectors beyond ad tech. They want to explore health care, life sciences and financial services, all of which also have a need for secure data collaboration technology.
Before Samooha, Bhowmick spent more than five years running machine-learning-based privacy and cryptography efforts at Apple, where he developed and deployed PETs across more than 1 billion iPhones, including secure multiparty compute and differential privacy.
Sivaramakrishnan and Bhowmick met through a mutual friend, and she liked the way that he described Apple’s approach to computational privacy and using “PETs as a foundational platform for first-party applications.”
“They democratized privacy so it was possible for anyone in the enterprise to have that capability,” Sivaramakrishnan said. “[Our thinking was] why not build that natively into clean rooms to make them even more secure and intuitive so that enterprises essentially have no excuse not to do the right thing with their customer data?”
