Paradigm shifts are hard. When you’re used to doing something a certain way, realizing that you can do it in a completely different way (and get similar results) is really hard to grasp. It can take anywhere from 18 days to almost a year to form a new habit. So it’s no surprise that many advertisers are expecting to have a difficult time with audience targeting, once third-party cookies go away. 

But what if there is a way to achieve the scale and accuracy of audience targeting in today’s cookie-based world, without third-party cookies or any form of alternative identifiers? To explain, here’s a deliberately oversimplified example of how to do audience targeting without cookies.

Deliberately oversimplified audience targeting example 

Let’s use a hypothetical campaign with very simple audience targeting criteria. A video game company is launching a new game and wants to target “gamers.” In the current cookie-based paradigm, a programmatic buyer could go into their buying platform, search for “gamer” and run the campaign across any of the hundreds of relevant gamer audiences that show up.

In the post-cookie paradigm, the programmatic buyer has a much harder task at hand. Without gamer audiences readily available, he/she will have to use a variety of identity-free signals and tactics. In the examples below, I’m going to make some generalizations for the sake of simplicity but will show how three identity-free signals — hour, designated market areas (DMAs), and website domains — can be used to accurately target the gamers audience.

Using HOUR as a signal

There are 24 hours in a day. If you had to guess, what are the times that video gamers are browsing the internet and most likely to be receptive to an ad for a video game? To keep things simple, let’s say that late nights from 9 p.m. to 2 a.m. are more likely than early mornings from 6 a.m. to 9 a.m.

Using DMA as a signal

There are 210 DMAs in the US. Using first-party data or third-party research, a brand can determine those that might be the best targets for a campaign.

Based on a study from WalletHub that used factors like internet speed and video-game stores per capita, certain cities in the US tend to be very conducive for a gamer lifestyle. These include Orlando, Seattle, Austin, New York, Los Angeles, Las Vegas, Irvine, Boston, and San Diego. This is just one data source, but for the purposes of this hypothetical campaign, it’s a good list of areas to target gamers.

Using DOMAIN as a signal

There are a handful of domains that gamers frequent to learn more about games:,,,, and But of course video gamers also visit websites not related to gaming at all. Tech blogs like and, music sites like and, news sites like and, and much more.

Letting AI do the complicated parts for audience targeting 

Using these hours, DMAs, and domains as individual signals would probably yield a decent strategy for targeting an audience of gamers. But what if I told you there is a way to get much more accurate results, with a lot less manual work?

Using AI, there are ways to identify the most precise hours, DMAs, and domains that work best for an audience without having to have any pre-existing understanding of that audience or having to perform market research to figure out the signals. Further, as the target audience changes and shifts over time, the AI model will pick up on nuanced changes in behavior and automatically adjust to pick out the most accurate signals.

Using AI, decisions can also be made not only across the individual signals but across a combination of all three signals for maximum precision and accuracy. For example, imagine if an ad opportunity at 10 p.m, in the Orlando area, on was available in the buying platform. There’s a really good chance that you’d be looking at a gamer.

This might sound simple, but it can get complicated really fast. Take 24 hours, 210 DMAs, and let’s just say 50,000 domains. That creates more than 250 million combinations that have to be scored and ranked to accurately target a gamer audience! And of course, the audience targeting strategies that brands deploy get much more granular and complex than simply targeting “gamers.” They want to target gamers who use specific consoles or have preferences in the kinds of games that they play.

Shifting to id-free signals

The good news is all this complication is a problem that machine-learning and AI have been solving in digital advertising for a decade. While the application to date has primarily been for a cookie-based world, reapplying this technology to identity-free signals — and shifting our preconceived notions on how audience targeting has to work — would create a path to scalable and performant audience targeting, without using any IDs at all.

If you want to learn more about Dstillery’s latest product offerings, contact us.