Deep Fake Series – Part 3
5 Rules to Test the Authenticity of In-Market Audience Data
I own 22 cars, speak six languages and am currently in-market to buy an additional 34 cars. This profile is associated with me via the third-party cookies on my computer. My profile, and millions more like it, are for sale today, from data marketplaces that sell in-market audience data. If you are currently using any form of “in-market” audiences, you are likely buying this sort of garbage. For the record, I own two cars, speak one language and I am not in-market for any cars.
The example I gave above falls under the category of in-market audiences based on modeled data, although they would like you to think it is behavioral data.
We’ve been making our way down the In-Market Audience Funnel. In its purest form, behavioral in-market audiences are segments that only contain behaviors from consumers who are actively shopping online within the last 30-60 days. These behaviors, or in-market signals, include consumers shopping on your website or shopping on an auto marketplace or auto review website. However, we find most often that most of these “in-market” audiences don’t contain any behavioral or in-market signal data. That is why in the above profile, I was in-market for 34 cars. These in-market segments were derived from models, not real-time behavioral data.
Here are 5 rules to live by when assessing the quality of in-market behavioral audiences:
Just because they have in-market data, doesn’t mean it’s accurate. Be skeptical and remember, the vast majority of in-market audience data is fake. Utilize the additional rules below to determine if you are buying fake in-market audiences versus authentic, real-time behavioral data.
How are the audiences created? Are the audiences created from models or real-time behavioral data? You need to read the description or fine print about the audience segment you’re buying so you understand the methodology used to create the audience segment. A clear indication that you are buying modelled audiences is to look for words like “modeled” or “derived from online behaviors” in the description. The latter description, “derived from online behaviors,” is another way to say “modeled.” You want to make sure that the in-market audience segments are composed entirely of online shopping behaviors within the last 30-60 days.
How much does it cost? This is the cost in addition to the cost of the media. Sometimes this is difficult to determine because the audience fees and media are bundled together. If your provider is promoting low cost CPM’s, they are most likely utilizing modeled or fake data. Real-time behavioral data is expensive, and if the quality is good, is almost always worth it.
What is the source of the in-market data? If they are buying in-market audience data from an audience data marketplace, the vast majority of this data is garbage because the higher quality, in-market data is typically never available for anyone to buy in the open marketplace.
Test for Performance. This is the ultimate test. However, it is difficult to measure. You should measure the performance of in-market audiences based upon sales impact, not arbitrary and vanity metrics like clicks, click-through rate etc. This requires doing a 1:1 exact sales match to the ad exposure or click. Very few companies are doing this today. As a result, ask for proof. Ask to see the ad exposure logs or the entire consumer purchase journey that are matched to the sale.
Next Up: First-Party Data
We are down to the last two sections of our In-Market Audience Funnel, dealer websites and CRM/DMS data. These two segments constitute “first-party data.” First-party data is a hot topic for marketers at the moment so don’t miss out! Stay tuned for our last installment, we will discuss the bottom of the funnel and guide you to create, buy or find your own in-market audiences using your dealers’ very own data.