“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 Joseph Lavan, vice president of data and insights at Netmining.
When trying to predict human behavior, there’s a quote I always look back on: “Sooner or later, everything old is new again.”
Think about it. One of the hottest shows in 2016 was the ’80s-style horror remake “Stranger Things.” There are scrappy entrepreneurs buying up retro Nintendo games and reselling them for massive profits on sites like eBay and Amazon. And at Mobile World Congress this week, all anyone could talk about was the return of the classic, early 2000s Nokia 3310 mobile phone.
So, how does this relate to digital marketing? The old is becoming new again when it comes to attribution measurement.
Test and control is an analysis method that has been used in advertising since the days of print and direct mail and is as “traditional” a technique as they come. Now it is emerging as perhaps the clearest way to understand which media channels are driving sales. This is fueled by the explosion of digital consumer touch points, granular audience targeting and the fact that the long-promised cross-marketing attribution golden goose has still not arrived.
For a digital ad industry that is still stuck using flawed last-touch attribution models, test and control could pave the way for more detailed understanding of digital media spend. Perhaps best of all, advertisers can leverage the tactic with data they already have.
When it comes to combatting the issues around attribution, there are two key sources of data: a brand’s CRM file and its transaction logs. CRM represents the starting point, and more and more brands are getting wise to its utility in online advertising. Onboarding a CRM file and matching it against another consumer data set, such as those provided by Acxiom and other companies, allows brands to identify and target existing customers online in an anonymous and PII-compliant fashion. This also lets brands simultaneously identify prospects who have similar attributes to their existing customers.
Armed with this understanding, brands can analyze their audiences to identify the highest-value customers and build marketing strategies that pursue these segments. Boiled down, this strategy becomes about targeting two pools: the consumers who appear to be the best possible prospects and the existing customers who represent the highest likelihood to spend more with the brand.
The next step is where we take it on back to the old school. Brands then need to take their two lists and identify test markets, such as physical retail locations or regions, and then pull a sample of households out of each list to serve as a control group. This group will never see online ads, nor will they receive any offline direct mail offers; campaigns like this should be done simultaneously for a clearer understanding of performance attribution.
Post-campaign, brands need to share their transaction files so that their agencies, data partners and ad platforms can analyze in-store and web transactions. Through a match-back process, these partners can understand which households saw an online ad and made a purchase, which saw a print ad and made a purchase, which saw both and which made no purchase at all.
Consumers grouped together in an audience segment do not all respond to the same tactics. Some will buy a product based on an email, others because they saw both a display ad and the email. When compared against the control group – which, remember, contains existing customers who are already likely to make purchases – this provides a much deeper understanding of how well each campaign component performed in driving sales. The goal is to find the consumers who deliver healthy ROI and paint a more robust picture of how media led those consumers to the sale.
Right now, tactics like this represent an important rebuttal against last-touch or last-click attribution. Nearly everyone in the industry understands that those methods are flawed, but they continue to use them because other attribution strategies are few and far between. It’s easy to distort last-touch models, which can become exercises in cookie bombing that irritate consumers and fail to give brands any sense of where to invest their budget.
As the ad industry continues to expand its reach across more screens and addressable media touch points, last-touch attribution will become less and less effective. Brands are increasing their investment, sure, but what brands really want is to spend more money on something that they can prove actually works. The complexity of the media mix should push advertisers to go beyond cookie-based audience targeting, making household-level matching much more important.