Identity: Too Big To Fail?

  • "Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.

    Today’s column is written by Nancy Marzouk, founder and CEO at MediaWallah.

    Without a doubt, identity is the capability du jour. It seems like identity is everywhere, and a slew of new identity players – plus more than a few legacy companies that have re-emerged as identity solutions – are riding the wave.

    But I can’t help but feel déjà vu. So many times our incredibly talented industry has solved for a significant problem, only to end up on the all-too-familiar path from exciting promise to oversold innovation.

    If we’re not careful, identity could meet a similar fate ­– and a similar backlash. We can’t afford to let that happen. Identity holds the potential to solve almost every major problem that marketers have faced, from omnichannel measurement to cross-channel targeting and the holy grail of one-to-one customer conversations at scale. It’s too big of an opportunity to squander on bubble thinking.

    We must address the ways we’re already oversimplifying – and overhyping – identity today. Here are three of the most egregious oversimplifications I regularly come across in conversations with identity consumers and providers alike.

    I hope my observations help wiser heads prevail.

    Deterministic data means trustworthy data

    Most, if not all, identity products use deterministic identity data to power them. By deterministic data, I mean identifiers that can be directly linked to the same person, such as an email address submitted by a consumer with an online purchase, the cookie that is dropped by the ecommerce site and then any other email addresses used by this now-cookied individual to sign in to other sites.

    Deterministic identity connects these and many other coinciding identifiers together, to pinpoint which signals identify the same person.

    It’s a common belief that deterministic identity products – whether they’re solely deterministic or rely heavily on a deterministic truth set – are inherently reliable since deterministic data is perceived to be accurate and precise. But this isn’t necessarily true.

    Since deterministic data builds layer upon layer of knowledge, any bad data in the mix can corrupt the subsequent layers of data. A wide array of factors – from human error to bot traffic – can and regularly do lead deterministic systems astray.

    Since deterministic data is so central to identity, solution providers and buyers alike must be vigilant about the deterministic “ingredients” that go into their identity products. It’s on all of us to ask tough questions about how data was gathered, how data sets were built and how any inaccuracies get weeded out.

    Identity data is ready to use

    Once the separate identifiers are gathered, many believe that simply normalizing the data – essentially, writing it in the right format – is the only step needed to match customer identities.

    Normalizing really is the first step before processing the data to weed out anomalies and bad data to make it fully prepared. Identity data is aggregated from many sources, which can use different formats and data types.. Different providers anonymize different users using different and often conflicting algorithms. And manually recorded information lives alongside automated feeds.

    That lack of standardization is a recipe for confusion and error. It’s on the identity provider to integrate different types of data, make rules for how to interpret conflicting data sets and decide when data is so unstandardized that it’s unusable.

    In that jumble of information, normalization alone isn’t enough. Identity data needs a robust standardization process. Otherwise, you’re just mixing signals.

    One size fits all

    Imagine two groups in a marketing department, both on the lookout for an identity solution. One, a direct marketing team running an integrated campaign, needs to identify new customers – perhaps hundreds of millions of them – to avoid over-messaging across platforms. The second is an analytics group, interested in granularly observing cross-channel engagement across a statistically relevant sample.

    What do these groups need from their identity offering? The direct marketing group will need impressive scale but requires only moderate accuracy. By contrast, the analytics group does not need massive scale but does need a high degree of accuracy to achieve its work.

    Too often, those signing off on the identity purchase will decide that one identity solution suffices for both teams. If all identity solutions were equally effective at handling all identity problems, that would make sense. But that’s not the reality. Providers have different strengths in handling factors such as scale, precision and accuracy. Brands need to understand what their use cases really call for – and which provider, or group of providers, can really deliver what they need.

    Honestly facing identity’s myths will take patience from and might even cause frustration in buyers and providers alike. But for identity to live up to its promise, we need this kind of honesty. Identity is an innovation that’s far too critical to squander – and it’s not too big to fail.

    Follow Nancy Marzouk (@nmarzouk) and AdExchanger (@adexchanger) on Twitter.

    Opinion > Data-Driven Thinking