Dotdash Meredith Sees Off-Platform Ad Targeting as Big Revenue Driver

Dotdash Meredith is testing out its D/Cipher ad intent targeting program off its own properties and on other sites across the internet. The results are looking good. 

Working in partnership with key clients, in January, D/Cipher+ started finding opportunities for ad placement that other technologies—namely cookies—were not finding. As it expands to non-DDM URLs, the company anticipates that the results will be very, very fruitful.

“D/Cipher Plus, in our partnership with OpenAI, and also on our own, we’ve mapped comparable third party sites that have the same signal of intent that D/Cipher utilizes from the signals developed on our own properties to target and deliver ads,” Chris Halpin, IAC chief financial and operating officer, said on the quarterly earnings call on February 12. “We can substantially increase our supply of impressions and do it at attractive margins, while providing our advertisers with exceptional performance.” 

We “believe it can grow rapidly on a revenue basis,” Halpin said.  

The technology, now powered by OpenAI, has been in the works for a decade and surpasses cookies as a tool to reach consumers, Jon Roberts, chief innovation officer at Dotdash, told AMO. 

“Orders that include D/Cipher… provide over half of the direct digital revenue in DDM,” Halpin said. “Orders with D/Cipher are over 50% larger than orders without D/Cipher.” 

Confused? Here’s a short cut. 

Third-party cookies have been the standard mechanism for tracking individuals across the internet for years, allowing advertisers to run ads convincing you to buy things days or weeks after you had already purchased them, making it possible for gift recipients to know what you got them ahead of time. Google was supposed to have gotten rid of them in Chrome, but kept kicking the can down the road until it just changed its mind. Nonetheless, DDM has been looking for alternatives for a long time. 

What D/Cipher does instead of looking at individual people, is it looks at the content they consume and how there may be crossover—in real time. For example, if you navigate to a page to read about whether a 100 degree Fahrenheit fever warrants taking your kid to the doctor, you’ll immediately get an ad for, say, Tylenol, because clearly at the moment in time you are parenting someone. You will not get an ad for a pair of jeans you bought last week—the cookie model. 

Cookies are backwards, while D/Cipher is forward-looking, Roberts told AMO. And you should know your audience and what they want at the time they are looking at a particular site. The fact that DDM has such a huge breadth of content has helped them see connections across topics that other brands might not be able to see.

“The thing that’s given us a real advantage, a step forward here is that because we have the great finance business, the great health business, we can actually see correlations,” Roberts said. Someone reading about Alzheimer’s is likely to also be interested in retirement homes and 401ks. You can track the overlap of people who read one story that then go to another and the more overlapping, the greater the score—that’s what D/Cipher, well, deciphers.

“We have that data, it’s very robust and very stable, so this allows us to take a really clear core investment, very simple to understand, looking at audience at how that behavioral expansion works,” Roberts said. “Again, we’re not retargeting these people.”

A key to the mission behind D/Cipher was also to take unsold ad inventory and put ads there. 

“The principle behind this was there is a premium inside publishing that’s not currently being accessed,” Roberts said. So D/Cipher can help find cheaper ads that they can sell at a higher price in places others have not yet discovered.  

An example of how OpenAI has helped crack this nut is that it has an understanding of nuances that earlier LLMs didn’t have. For example, OpenAI can recognize synonyms, where old technology couldn’t. That is a fresh new opportunity to create connections and understand where people are going. Or, just because you’re on a page about a museum in Hamburg, doesn’t mean you care about museums in Boston. You’d probably, however, care more about restaurants and hotels in Hamburg. AI can make those distinctions. 

“You’re going to have more value in that audience than is captured by the market. So the goal here is bring a premium back to the underbought ads,” he said. It’s a business DDM got into because they saw that gap in the market with undersold inventory—and it’s not something they would have done for a couple million dollars of additional revenue. “We think this will be a material business.” 

How the ads will show up on third party sites and the technical aspects of the business remain to be seen.