Deep Fake Series - Part 1

In-market audience segments make big promises of hyper-targeted marketing. However, not all in-market audiences are created equally. We refer to some as deep fakes. In case you’re not familiar with the term deep fake, merriam-webster.com defines deep fake as referring to “a video that has been edited using an algorithm to replace the person in the original video with someone else (especially a public figure) in a way that makes the video look authentic.”

Deep fakes undermine our ability to trust what we see and what we know. Probably the most well-known deep fake is Tom Cruise on TikTok .

At Clarivoy, we use the term deep fake to refer to in-market audience segments that are, well, fake. They don’t actually contain many in-market shoppers at all.  In fact, we discovered one data broker that sells an audience segment called “in-market Toyota Highlander shoppers,” which contains 58 million in-market Toyota Highlander shoppers. Clearly, that’s an outrageously inflated number!

Marketers need to understand how their in-market audience data set is being built so that they can evaluate the quality of the audience segment. In-market audience segments are not created equally, and no one wants to waste money on audience segments that aren’t targeting actual, in-market auto shoppers.

So, what is a marketer to do? How do you know what is fake and what isn’t? Well, we have a lot of information to help you sniff out the differences between authentic in-market audience segments and fake in-market audience segments. We have so much that we made this a three-part series.

First, we created a framework, called the In-Market Audience Funnel, to help better explain the various types of in-market audiences. The layers of the funnel represent various types of in-market audiences available today, from “modeled” at the top all the way down to dealers’ websites and CRM/DMS data at the very bottom. In this article, we will look at modeled audiences along with audiences created using credit data.

Modeled audience segments contain little to no real-time behavioral shopping data, and often are based on random and universal data points like whether or not a person uses a credit card. (How many people do you know who don’t use a credit card?!?) When we dug deep into modeled audiences, other data points used to create the audience segments were: age, income, interests, whether you used the internet or not, and even whether or not you frequented stationery stores. Yes, you read that correctly. One in-market audience data provider uses factors such as whether you used the internet and frequented stationery stores to determine whether you are an in-market shopper.

Modeled audiences are based on such broad and random criteria that putting our faith in their authenticity requires naivety and foolish optimism. Modeled audiences could even be called “guessed” audiences because really they are created by guessing, and connecting random criteria.

To better illustrate this, I utilized an online tool provided by a data marketplace that will show you all the audience segments with which you are associated. The tool scans the third-party cookies on your computer and then generates a report. The report identified me as an in-market shopper for EVERY segment and make in the auto industry: Ford, Honda, Chevrolet, Mazda, Toyota, full-size truck, small cross-over, EV, minivan, compact sedan, etc. The company also showed me as owning 22 different makes of vehicles, some of which aren’t even sold in the U.S. I am not in the market for any of the makes or segments they listed, and I only own two cars, but my cookie and identity are being sold to dealers as though I am an in-market shopper owning 22 different makes of vehicles. Modeled audiences do have a benefit: they are cheap. However, as the saying goes, you get what you pay for.

The next portion of the funnel is credit data. These in-market audiences are created by purchasing consumer credit data directly from companies such as Equifax. Be careful though, there are credit audiences sold on the open market that are modeled. Be sure you are buying live credit data, which requires, among other things, that the consumer receives a firm offer of credit as part of the marketing campaign. You can create highly targeted lists for use in your digital and programmatic campaigns so long as you remain compliant with laws governing credit data. For example, you can create a list of everyone in a specific zip code with a credit score greater than 700 who has a lease expiring in the next 120 days. Your credit audience segmentation options are only limited by your imagination.  

Understanding in-market audience options – their costs and benefits- takes time. Our in-market audience funnel is a lot to digest. So, go back to the top, review, and keep your eyes open. Coming soon is part two in our deep fake series in which we deep dive into contextual and behavioral in-market audiences. Don’t miss it!

‍