How The New York Times mines data to pick articles to promote on Facebook and Twitter

“In our test, it’s shown to basically perform at half the [cost per order] of a manual campaign,” said Colin Russel, senior data scientist at the Times who created the tool.

The Times no longer uses Facebook and Twitter pixels on most of its pages. Chris Wiggins, the Times’ chief data scientist, said that instead of using a reader’s browser history to gain more insight into the types of content they tend to read, TAFI looks at how an article has been performing organically on social and factors that into the data the Times has on readers’ interests. 

For the past decade, tracking pixels on social media platforms was believed to be the only way for a marketer to achieve an effective and intelligent marketing campaign. 

“We’ve been able to show that we can use our data without that sort of exposure of reader history via a tracking pixel,” said Wiggins. 

Russel said that tracking pixels have been removed from all articles and browsing pages, though certain pages on the site still have them. Russel declined to share which ones.

“TAFI efficiently spends money on Facebook and Twitter, so they’re certainly benefiting because it allows us to achieve a very low CPO, which encourages us to use those platforms more,” said Russel. “We can get more bang for our buck.”

And because of the improvements in efficiency of tracking how well a campaign is performing, it also changes the calculations on how much money should go to these platforms overall, because the marketing team knows it’s not wasting money.

Russel said that the idea for the tool came around two years ago and was first launched to run international campaigns a year and a half ago. Then, a year ago, the Times pulled out of their third-party subscription marketing partnerships. 

The goal for the near future is to use the homegrown tool on other platforms, such as Snapchat and Reddit, along with paid search and paid display advertising.