The number of alternative IDs in the market is daunting. But OMD has a philosophy for determining which IDs will work best for brands’ needs.
Emily Proctor, the ad agency’s executive director of data and technology solutions, shared that philosophy at AdExchanger’s Programmatic IO in Las Vegas on Tuesday. It boils down to this: “Test, learn, scale, optimize, repeat.”
That ethos applies to testing which alt IDs to adopt before scaling up spend and continuously evaluating those IDs post-adoption, Proctor said. And, she emphasized, marketers need to use a mix of IDs tailored to brands’ specific use cases.
“There’s no silver bullet,” she said. “There never will be.”
During her Programmatic IO presentation, Proctor outlined the criteria OMD uses to evaluate alt IDs. She also shared which five IDs OMD uses most often and why.
Key criteria
OMD weighs five factors when testing alternative IDs, Proctor said: scale, accuracy, interoperability, privacy and implementation.
Rather than seeking scale for scale’s sake, consider first and foremost whether the alternative ID will help the brand reach its specific audience, Proctor said. Most likely, achieving the ideal scale will require adopting multiple IDs for different media channels.
“What works in one environment – say, CTV – doesn’t necessarily translate into the open web or in-app,” she said. “You really need to have a diversified identity approach tailored to your specific media mix.”
Scale is also tied to accuracy, she said: Consider how high the audience match rates are for the ID and how precise the audience data is that underlies that ID. First-party data sets tend to be a better foundation than third-party connections, she added.
On the interoperability front, consider whether the alternative ID has integrations across your DSPs and data partners, Proctor said. Also weigh whether the ID is in compliance with data privacy standards and whether the underlying data was collected in a way that respects consumer consent.
And, finally, evaluate how easy the ID is to integrate and activate within the advertiser’s tech stack and whether the results are worth the effort involved.
“There’s a fair amount of half-baked solutions still being tested out in the marketplace, and not every alt ID is going to be right for your business,” Proctor said. “So you need to ask some tough questions: Is this a product, or is it an idea?”
OMD has also found AI helpful for making sense of the various data signals from different alt ID providers. For example, Proctor pointed to a large global CPG client that is currently working with OMD to test solutions across different markets.
Because no single ID provider can own the aggregation of that customer data across those markets, OMD is using its Omni Assist AI agent to aggregate the data instead.
AI can “analyze those results at a much faster pace than a human could do it,” Proctor said. However, she added, while AI can get you “80% of the way there,” humans need to verify the AI findings.
OMD’s fab five
After testing alt IDs across multiple brand clients, OMD has landed on five identifiers that it uses most often: Google’s ad ID, Yahoo’s ad ID, Neustar’s Fabrick, LiveRamp’s RampID and UID2, which is managed by The Trade Desk.
All of these IDs perform well against OMD’s evaluation criteria, Proctor said. But she pointed to a fundamental distinction between them.
The Google and Yahoo IDs are owned by walled garden platforms, which means advertisers have to use them in order to get the most out of audience targeting and measurement on these platforms, she said.
Meanwhile, Fabrick, RampID and UID2 are all open identifiers that prioritize interoperability and privacy compliance, Proctor said. They also offer scale without being locked to a single platform.
While these five IDs are OMD’s “across the board” recommendations, she said, “if you get down to the client level, those might not be the ones that we recommend, because it really depends on what their strategy looks like, what their media mix looks like and what their goals are.”
UID2, emphasis on “too”
Proctor also answered a few audience questions about UID2 in particular and why The Trade Desk’s endorsement doesn’t make it a hands-down winner.
UID2s can perform extremely well if a brand has a strong first-party data pool, Proctor said. She pointed to a recent campaign OMD ran for McDonald’s in Australia that aimed to bring two million new users into the fast-food chain’s app-based loyalty program.
OMD used UID2s from existing McDonald’s loyalty program users to create lookalike models for new users to target. Rather than solely relying on a lookalike model that weighs all the characteristics loyalty program users have in common, Proctor said, McDonald’s extensive data from in-app purchases let it get more granular. For example, it could consider whether consumers tend to purchase more breakfast orders or whether they’re afternoon coffee drinkers, and then target the campaign accordingly.
Compared to a third-party cookie-based strategy, the UID2-based campaign produced a 92% revenue lift, 87% increase in purchases, 66% lift in click-through rate and 31% higher return on ad spend.
Those results owe a lot to McDonald’s first-party data strategy and the robustness of the data it has on its users, Proctor said. But not every brand has that data foundation, which is why OMD recommends building one.
So, although UID2s perform, and the ID has the backing of The Trade Desk – the industry’s largest independent DSP – that doesn’t make it the end all be all, Proctor said.
Plus, not every brand uses The Trade Desk for its campaigns, she said. And many that do use the DSP don’t invest in the platform at the same level as others, so putting all their eggs in the UID2 basket doesn’t make sense, she added.
Choosing alt IDs “really has to be a bespoke strategy,” she said, based on “what your clients are trying to accomplish.”
