Skip to Content

Ads That Follow Customers Around the Web

Algorithms distinguish casual website visitors from people who truly are potential customers, so the likeliest buyers can be shown ads while they surf.

For every 100 people who visit a retailer’s website, only about two will buy something; the other 98 will just leave. But with enough reminders, some of these 98 people might become customers someday. The trick is to figure out which ones they are, and then to nudge them by means of targeted ads. The practice is known as retargeting, and it’s growing on the Web.

Remember this?: Someone who browses listings for bike trailers on a site run by CSN Stores (upper left) might later see a related ad on a separate website, such as Weather Underground.

When you see an online ad for a product you previously looked at on a different website, you’ve been retargeted. Making that happen is the business of companies like TellApart, which was founded in 2009 by Mark Ayzenshtat and Josh McFarland. They came from Google, where they had both worked on AdWords and other advertising platforms. TellApart uses data it gleans from online retail traffic to predict whether someone who leaves a given website is still a likely customer of that site—or not. People deemed likely to return can be shown ads for the site whenever an opportunity arises on the real-time bidding exchanges used to fill online ad space. “We only get paid when we drive a sale,” says Ayzenshtat. “We’re confident that this works.” About 7.5 percent of viewers click through on their personalized ads, the company says. If so, that is dozens of times higher than the click-through rates on generic Web ads. And about 4.5 percent of those who click on a TellApart-generated ad eventually make a purchase, the company says.

Ayzenshtat says the company’s algorithms assess site visitors much as a salesperson would in a physical store. “When you walk in, the salesperson develops a profile based on what you’re wearing, who you’re with, if you have other shopping bags,” says Ayzenshtat. “In the online world, the signals are harder to tease out.”

To do it, TellApart’s servers sift through a tremendous amount of data provided by the online retailers it works with. For instance, when someone visits a site, it can examine where the person had been previously: did he click a link from a blog or an ad? Did he end up at Diapers.com after searching for a particular crib that he had also checked out on other sites? And once on the retail site, what did he do? Did he leave after seeing one page, or did he seem to be comparison-shopping? Also factored in are what products a customer browses on the site, whether they are all in a particular category, and how popular these items are with other visitors. TellApart’s algorithms use the time of day and location to determine whether a person is browsing at home or at work. It follows the user by placing a cookie in his Web browser, a standard method of keeping track of people online.

TellApart’s customers include many of the Web’s biggest retailers, including CSN Stores, which operates the largest online-only home goods store through a network of more than 200 sites including Luggage.com and Cookware.com. The company had $380 million in sales last year and is expanding internationally. Jeff Steeves, director of customer acquisition at the company, says it has worked with TellApart since last May. “We see a significant lift to our conversion rate,” he says. “They do the heavy lifting for us, operating models to measure a user’s engagement and determine … which are most likely to make a purchase if shown a targeted ad.” The company doesn’t want to irritate unlikely buyers or spend money on ads that are unlikely to lead to a sale. “Our breed of remarketing is subtle,” says Ayzenshtat.

All this information gives TellApart a very detailed picture of a site visitor. The retailer does not, however, disclose to TellApart a person’s name or username. “We’re not interested in that,” says Ayzenshtat. The CEO of another retargeting company, who did not want to be named for this story, noted that user privacy is mostly self-regulated by the big players in the online ad industry, especially Google—and retargeters would like to keep it that way rather than have the U.S. Federal Trade Commission, for example, get involved. One of the principles of this self-regulation is that users have to be able to opt out of an advertising technology. TellApart satisfies that requirement by putting a little “x” in the corner of ads it shows, though it says only a tiny fraction of the people it targets have opted out.

Retargeting is likely to get even more precise. Ayzenshtat says TellApart has determined that it can drive a 30 percent increase in sales by picking out which consumers are more likely to buy something when presented with a coupon, and which ones don’t need coupons as an inducement. Similarly, some sites offer an opportunity to put in a coupon code when checking out; some people who don’t have one will simply ignore it and follow through on a purchase, but others will go out and search the Web for mentions of that code, find it, and then return to complete their transaction with a discount of, say, 20 percent. Now TellApart is developing a way for retailers to know when to display a place to enter a coupon code, and when they don’t need to bother.

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

This baby with a head camera helped teach an AI how kids learn language

A neural network trained on the experiences of a single young child managed to learn one of the core components of language: how to match words to the objects they represent.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.