A Map or a Matrix? Identity Management Is More Complex By the Day

A newly published white paper on how advertisers and brands can recognize unique customers across marketing platforms underscores just how tough this important job is for data-driven marketers.

As technologists and policymakers weigh in themselves on the data universe – often without understanding the full ramifications of what they do (or worse, knowing so but proceeding anyway) – data flows on the Internet and on mobile platforms are being dammed, diverted, denuded, and divided.

In my opinion, these developments are not decidedly good for advertising – which relies on such data to deliver relevance in messaging, as well as attribution and measurement. There is a troubling anti-competition mood in the air. It needs to be reckoned with.

Consider these recent developments:

  • Last week, the European Court of Justice rendered a decision that overturned “Privacy Shield” – the safe harbor program that upward of 5,000 companies rely upon to move data securely between the European Union and the United States. Perhaps we can blame U.S. government surveillance practices made known by Edward Snowden, but the impact will undermine hugely practical, beneficial, and benign uses of data – including for such laudable aims as identity management, and associated advertising and marketing uses.
  • Apple announced it will mandate an “opt-in” for mobile identification data used for advertising and marketing beginning with iOS 14. Apple may report this is about privacy, but it is also a business decision to keep Apple user data from other large digital companies. How can effective cross-app advertising survive (and be measured) when opt-in rates are tiny? What about the long-tail and diversity of content that such advertising finances?
  • Google’s announcement that it plans to cease third-party cookies – as Safari and Mozilla have already done – in two years’ time (six months and ticking) is another erosion on data monetization used for advertising. At least Google is making a full-on attempt to work with industry stakeholders (Privacy Sandbox) to replace cookies with something else yet to be formulated. All the same, ad tech is getting nervous.
  • California’s Attorney General – in promulgating regulation in conjunction with the enforcement of the California Consumer Privacy Act (in itself an upset of a uniform national market for data flows, and an undermining of interstate commerce) – came forth with a new obligation that is absent from the law, but asked for by privacy advocates: Companies will be required to honor a browser’s global default signals for data collection used for advertising, potentially interfering with a consumer’s own choice in the matter. It’s the Do Not Track debate all over again, with a decision by fiat.

These external realities for identity are only part of the complexity. Mind you, I haven’t even explored here the volume, variety, and velocity of data that make data collection, integration, analysis, and application by advertisers both vital and difficult to do. As consumers engage with brands on a seemingly ever-widening number of media channels and data platforms, there’s nothing simple about it. No wonder Scott Brinker’s Mar Tech artwork is becoming more and more an exercise in pointillism.

Searching for a Post-Cookie Blueprint

So it is in this flurry (or fury) of policy developments that the Winterberry Group issued its most recent paper, “Identity Outlook 2020: The Evolution of Identity in a Privacy-First, Post-Cookie World.”

Its authors take a more positive view of recent trends – reflecting perhaps a resolve that the private sector will seize the moment:

“We believe that regulation and cookie deprecation are a positive for the future health and next stage of growth for the advertising and marketing industry as they are appropriate catalysts for change in an increasingly privacy-aware consumer environment,” write authors Bruce Biegel, Charles Ping, and Michael Harrison, all of whom are with the Winterberry Group.

The researchers report five emerging identity management processes, each with its own regulatory risk. Brands may pursue any one or combination of these methodologies:

  • “A proprietary ID based on authenticated first-party data where the brand or media owner has established a unique ID for use on their owned properties and for matching with partners either directly or through privacy safe environments (e.g.: Facebook, Google, Amazon).
  • “A common ID based on a first-party data match to a PII- [personally identifiable information] based reference data set in order to enable scale across media providers while maintaining high levels of accuracy.
  • “A common ID based on a first-party data match to a third-party, PII-based reference data set in order to enable scale across media providers while maintaining high levels of accuracy; leverages a deterministic approach, with probabilistic matching to increase reach.
  • “A second-party data environment based on clean environments with anonymous ID linking to allow privacy safe data partnerships to be created.
  • “A household ID based on IP address and geographic match.”

The authors offer a chart that highlights some of the regulatory risks with each approach.

“As a result of the diversity of requirements across the three ecosystems (personalization, programmatic and ATV [advanced television]) the conclusion that Winterberry Group draws from the market is that multiple identity solutions will be required and continue to evolve in parallel. To achieve the goals of consumer engagement and customer acquisition marketers will seek to apply a blend of approaches based on the availability of privacy-compliant identifiers and the suitability of the approach for specific channels and touchpoints.”

A blend of approaches? Looks like I’ll need a navigator as well as the map. As one of the six key takeaways, the report authors write:

“Talent gaps, not tech gaps: One of the issues holding the market back is the lack of focus in the brand/agency model that is dedicated to understanding the variety of privacy-compliant identity options. We expect that the increased market complexity in identity will require Chief Data Officers to expand their roles and place themselves at the center of efforts to reduce the media silos that separate paid, earned and owned use cases. The development of talent that overlaps marketing/advertising strategy, data/data science and data privacy will be more critical in the post-cookie, privacy-regulated market than ever before.”

There’s much more in the research to explore than one blog post – so do your data prowess a favor and download the full report here.

And let’s keep the competition concerns open and continuing. There’s more at stake here than simply a broken customer identity or the receipt of an irrelevant ad.

Consumer Marketers, Looking to Test New Data Categories? Try These

We are all trying to create and sustain customers, using data to discover new patterns, new audiences, and new prospects — and that requires a lot of testing, and innovative data sets to explore (responsibly). Let’s make it experiential, as well as experimental.

We in the data marketing business love to test — at least, we should. And what we should test for is new data categories.

Expanding the marketing universe — and stretching the marketing budget — depends on higher efficiency in our lists, offers, and creative. We should be eager to test new proofs of concepts and new categories of data sources as they enter the market … if only to know whether or not they produce incrementally or otherwise.

I’m still surprised when I hear some of my data-vendor friends say that a good number of their clients pass on testing — and just go all-in on new lists and data sources. It seems like testing is still too much work for some, or they feel the only way to test is with an entire data source. Guess these client-side folks have money to burn, or are operating very much on-the-fly.

In some ways, digital marketers have it all over offline marketers in their ability to test, cycle, test again, and so on — often, many times over by the time a direct mail or direct-response print or broadcast test cycle has run its course. Yet, in this speed, have we sacrificed some quality in our prospecting strategies?

Online audience algorithms can produce some highly categorized niche segments, based on site visits and app usage — much of it de-identified, from a personal perspective. But how do these segments really stack up against a transaction database, or response lists, or even compiled lists, based on personally identifiable information? Thankfully, we can test for this, or even overlay data! (I am not advocating re-identification here, nor should you. Oh California, please don’t force us to identify non-PII. It’s soooo anti-privacy.)

Recently, the Direct Marketing Club of New York (DMCNY) held a very interesting breakfast program titled “Beyond Demographics: The Data You Need to Max Out Marketing Performance.”

Some Fresh Categories for New Reach and Affinity Discovery

Consider some of these data sources for testing:

  • Values Data — Test cohorts based on “shared values,” rather than simply choosing audiences based on demographics or psychographics. David Allison, principal, David Allison Inc., and author of “We Are All the Same Age Now,” pointed to his firm’s internal research that shows that popularly defined age groups rarely (or barely) match on what they agree upon, or value, as a generation. For example, Baby Boomers agree with each other about 13% of the time; Gen X, about 11% of the time; and Millennials, 15% of the time. Thus, targeting based on demographics alone can be extremely wasteful if the marketer is assuming some sort of shared attribute among them, other than age.However, when targeting based on shared “values” — Adventurers, Savers, and Techsters, and the like — all of a sudden affinities jump sky-high. In these cases, 89%, 76%, and 81%, respectively. These “valuegraphics” are based on “big data” segments — rather than small data (response lists, for example). Still, when compared to demographics targeting alone, shared-value targeting offers an eight-time lift!  Well, that’s worth testing.
  • Attitudinal Data — Another perspective on “beyond demographics” came from Mark Himmelsbach, co-founder, Episode Four, a creator of “brand hits,” such as this one for Charles Schwab. We often have stereotypical views of many demographic and other audience categories — and too many algorithms, he said. But analyze the data for unusual patterns, and suddenly you can find “who knew?” commonalities among certain audience segments that would wow any of us.Who knew that ultra-high net worth individuals are electronic dance music enthusiasts? Who knew that African-American married women are high on the e-sports genre? Or that young Hispanic/Latino adventurers are really into escape rooms? These discoveries give brands new advertising, product placement, and sponsorship opportunities, for example, which might otherwise go untapped. I’m still trying to get my head around these reported affinities, based no doubt by my own preconceptions.
  • Location Data — According to the World Economic Forum, 90% of the world will soon have or already has a supercomputer in their pocket — a smartphone. We’re actually closing in on four connected devices per person, reports Jeff White, founder and CEO, Gravy Analytics. With smartphones alone, as constant companions, we have a huge opportunity to leverage responsibly use of location data. Location can provide huge “affinity” targeting opportunities.A casual wine user might search and buy online his or her wine. But a wine aficionado visits a winery (Location X), or attends a wine tasting (Location Y), and now you have a true affinity opportunity. Granted, location data has a level of sensitivity that carries, more often than not, an opt-in requirement — but the marketing lift can be a significant reward for the advertiser who strategically applies such insights from it. Makes me want to tag every latitude and longitude for some hobby or interest!
  • Experiential Data — Live Nation may own concert venues, Ticketmaster, online game communities and music/culture festivals — but across these many first-party experiences, the company can provide deep analytics that help monetize its various audiences through enriched second-party relationships, said Anubhav Mehrotra, VP, Live Nation. Hilton, American Express, and Uber are just some of the brands Live Nation has teamed up with to enrich brand users with engaging experiences, such as backstage tours and “meet the artists.”

We are all trying to create and sustain customers, using data to discover new patterns, new audiences, and new prospects — and that requires a lot of testing, and innovative data sets to explore (responsibly). Let’s make it experiential, as well as experimental: I sure hope to meet some ultra-high-net-worth individuals at the next Electronic Dance Festival I attend. Or not.