main

ARTICLE

Don’t Be Like Biff in Your Approach to Data Management

March 28, 2017 — by John Slocum

What if I said you could know the future, just like the fictional character Biff Tannen from Back to the Future? When Biff stole a time machine and brought future knowledge back to his younger self, he was unknowingly flirting with “a time paradox, the result of which could cause a chain reaction that would unravel the very fabric of the space-time continuum and destroy the entire universe!”

It’s 2017, and now you can anticipate events and outcomes without time machines or unraveling the space-time continuum. In fact, with the state of audience management tools today, it’ll soon be irresponsible to spend your (or particularly your clients’) ad budget without already knowing the outcome. Balderdash, you say? Raving, frothing-at-the-mouth ramblings of a madman, you say?

Thanks to massive device and addressability proliferation coupled with user-level identification and the technology to manage said data, we can now ask and answer questions that predict audience response—before spending a dime of new budget.

Sound amazing? Just nod. I’m sure it sounds amazing. Instead of waving my hand and telling you, “These aren’t the droids you’re looking for,” let’s see what’s behind that statement.

Staying one step ahead of your customers

Workstations, laptops, phones, tablets, smart TVs, offline data—you name it, a modern data management platform can track it or ingest it. But it goes beyond device. To anticipate what your customers and prospects might consider next, you need user-level identification. You need access to human behaviors and outcomes. If you’re looking for performance across audiences, for example, you need to be able to sample those audience behaviors to predict what they’ll do next. Unlike the stock market, past audience performance IS indicative of future results!

This is a simple concept with less simple implementation. We’re talking about identifying the humans behind all those devices to analyze and predict what they’ll do next. Tablets don’t buy new shoes and handbags—my wife does. Piloting the technology to ask and answer the critical questions that predict audience response is the fun part, in my opinion.

There are many different approaches to predicting response. A couple key questions before we go shopping for ‘big data’ solutions—who is doing the predicting? Are you predicting results for your budget today? Are you predicting results for someone else’s budget? Is someone else predicting results for your budget? How are predictions measured? How far off do they need to be before budget, tools or even partners are reconsidered? Is the transparency available to confirm predictions? Can you predict performance for an audience and, if so, can you verify it?

Enter the modern DMP

Modern DMPs should be able to answer most of these questions for you, particularly around telling you how an audience will perform and which audiences will perform (or not). A modern DMP should be able to take your recent golf equipment campaign data, per human, and tell you how any group of humans in your addressable population actually performed in that campaign. If you have defined audiences, be they modeled, first-party site audiences or third-party audiences, you can sample overlaps of the campaign audience and your available audiences, calculating overlap performance (test group) versus the rest of the campaign audience (control group). Starting to sound a little like your Intro to Stats class? It should, these are basic concepts. We can calculate significance on these samples as well to determine how likely your sample represents the true population. Find audiences that are popping without spending on ‘em!

A modern DMP can also test performance of single audiences in real-time, upon creation, without new spending! This can be more of an ad-hoc, iterative approach—not testing means you are leaving insights on the table. This capability requires user-level event data, campaign data and a dynamic audience platform to make it sing; the one thing it doesn’t need is SQL! Not to worry, (fellow) data geeks, user-level campaign and audience data in a modern DMP is liberated, so you can grab that and get your hands dirty.  Many platforms are also offering user-level data in a self-service analytics platform to support your own ad-hoc queries, reports and even scheduled jobs.

You should know if your audience can deliver desired outcomes, before you spend. You should also be able to see exactly how, where and why it did, as it happens. Time travel is no longer required. Don’t be like Biff.