How data clean rooms might help keep the internet open
The IAB sees encouraging signs that DCRs might sustain addressable advertising outside the walled gardens and without the help of cookies.
Are data clean rooms the solution to what IAB CEO David Cohen has called the “slow-motion train wreck” of addressability? Voices at the IAB will tell you that they have a big role to play.
“The issue with addressability is that once cookies go away, and with the loss of identifiers, about 80% of the addressable market will become unknown audiences which is why there is a need for privacy-centric consent and a better consent-value exchange,” said Jeffrey Bustos, VP, measurement, addressability and data at the IAB.
“Everyone’s talking about first-party data, and it is very valuable,” he explained, “but most publishers who don’t have sign-on, they have about 3 to 10% of their readership’s first-party data.” First-party data, from the perspective of advertisers who want to reach relevant and audiences, and publishers who want to offer valuable inventory, just isn’t enough.
Why we care. Two years ago, who was talking about data clean rooms? The surge of interest is recent and significant, according to the IAB. DCRs have the potential, at least, to keep brands in touch with their audiences on the open internet; to maintain viability for publishers’ inventories; and to provide sophisticated measurement capabilities.
How data clean rooms can help. DCRs are a type of privacy-enhancing technology that allows data owners (including brands and publishers) to share customer first-party data in a privacy-compliant way. Clean rooms are secure spaces where first-party data from a number of sources can be resolved to the same customer’s profile while that profile remains anonymized.
In other words, a DCR is a kind of Switzerland — a space where a truce is called on competition while first-party data is enriched without compromising privacy.
“The value of a data clean room is that a publisher is able to collaborate with a brand across both their data sources and the brand is able to understand audience behavior,” said Bustos. For example, a brand selling eye-glasses might know nothing about their customers except basic transactional data — and that they wear glasses. Matching profiles with a publisher’s behavioral data provides enrichment.
“If you’re able to understand behavioral context, you’re able to understand what your customers are reading, what they’re interested in, what their hobbies are,” said Bustos. Armed with those insights, a brand has a better idea of what kind of content they want to advertise against.
The publisher does need to have a certain level of first-party data for the matching to take place, even if it doesn’t have a universal requirement for sign-ins like The New York Times. A publisher may be able to match only a small percentage of the eye-glass vendor’s customers, but if they like reading the sports and arts sections, at least that gives some directional guidance as to what audience the vendor should target.
Dig deeper: Why we care about data clean rooms
What counts as good matching? In its “State of Data 2023” report, which focuses almost exclusively on data clean rooms, concern is expressed that DCR efficacy might be threatened by poor match rates. Average match rates hover around 50% (less for some types of DCR).
Bustos is keen to put this into context. “When you are matching data from a cookie perspective, match rates are usually about 70-ish percent,” he said, so 50% isn’t terrible, although there’s room for improvement.
One obstacle is a persistent lack of interoperability between identity solutions — although it does exist; LiveRamp’s RampID is interoperable, for example, with The Trade Desk’s UID2.
Nevertheless, said Bustos, “it’s incredibly difficult for publishers. They have a bunch of identity pixels firing for all these different things. You don’t know which identity provider to use. Definitely a long road ahead to make sure there’s interoperability.”
Maintaining an open internet. If DCRs can contribute to solving the addressability problem they will also contribute to the challenge of keeping the internet open. Walled gardens like Facebook do have rich troves of first-party and behavioral data; brands can access those audiences, but with very limited visibility into them.
“The reason CTV is a really valuable proposition for advertisers is that you are able to identify the user 1:1 which is really powerful,” Bustos said. “Your standard news or editorial publisher doesn’t have that. I mean, the New York Times has moved to that and it’s been incredibly successful for them.” In order to compete with the walled gardens and streaming services, publishers need to offer some degree of addressability — and without relying on cookies.
But DCRs are a heavy lift. Data maturity is an important qualification for getting the most out of a DCR. The IAB report shows that, of the brands evaluating or using DCRs, over 70% have other data-related technologies like CDPs and DMPs.
“If you want a data clean room,” Bustos explained, “there are a lot of other technological solutions you have to have in place before. You need to make sure you have strong data assets.” He also recommends starting out by asking what you want to achieve, not what technology would be nice to have. “The first question is, what do you want to accomplish? You may not need a DCR. ‘I want to do this,’ then see what tools would get you to that.”
Understand also that implementation is going to require talent. “It is a demanding project in terms of the set-up,” said Bustos, “and there’s been significant growth in consulting companies and agencies helping set up these data clean rooms. You do need a lot of people, so it’s more efficient to hire outside help for the set up, and then just have a maintenance crew in-house.”
Underuse of measurement capabilities. One key finding in the IAB’s research is that DCR users are exploiting the audience matching capabilities much more than realizing the potential for measurement and attribution. “You need very strong data scientists and engineers to build advanced models,” Bustos said.
“A lot of brands that look into this say, ‘I want to be able to do a predictive analysis of my high lifetime value customers that are going to buy in the next 90 days.’ Or ‘I want to be able to measure which channels are driving the most incremental lift.’ It’s very complex analyses they want to do; but they don’t really have a reason as to why. What is the point? Understand your outcome and develop a sequential data strategy.”
Trying to understand incremental lift from your marketing can take a long time, he warned. “But you can easily do a reach and frequency and overlap analysis.” That will identify wasted investment in channels and as a by-product suggest where incremental lift is occurring. “There’s a need for companies to know what they want, identify what the outcome is, and then there are steps that are going to get you there. That’s also going to help to prove out ROI.”
Get MarTech! Daily. Free. In your inbox.