Amazon Knows What You Buy. And It’s Building a Big Ad Business From It.

SEATTLE — When a chain of physical therapy centers wanted new patients, it aimed online ads at people near its offices who had bought knee braces recently on Amazon.

When a financial services provider wanted to promote its retirement advisory business, it directed ads to people in their 40s and 50s who had recently ordered a personal finance book from Amazon.

And when a major credit card company wanted new customers, it targeted people who used cards from other banks on the retail site.

The advertisers found those people by using Amazon’s advertising services, which leverage what the company knows better than anyone: consumers’ online buying habits.

Many of Amazon’s features are similar to those of Google or Facebook, like offering ways to target users based on their interests, searches and demographics. But Amazon’s ad system can also remove a lot of the guesswork by showing ads to people who have bought the shirts on Amazon.com.

Advertisers have long run some targeted campaigns through Amazon’s ad network. Many have done that by working directly with Amazon’s staff, who would place their orders on their behalf. That option has historically been focused on larger brands because it requires a minimum advertising commitment. Over time, Amazon has given more advertisers and their agencies access to the self-service system to run their own targeting campaigns on and off Amazon’s websites, and at a variety of spending levels.

Users of the self-service system can choose from hundreds of automated audience segments. Some of Amazon’s targeting capabilities are dependent on shopping behaviors, such as “International Market Grocery Shopper” and people who have bought “Acne Treatments” in the past month, or household demographics, such as “Presence of children aged 4-6.” Others are based on the media people consume on Amazon, such as “Denzel Washington Fans” or people who have recently streamed fitness and exercise videos on Amazon. The company declined to comment.

Just the Cheese, a brand run by Specialty Cheese Company in Reeseville, Wis., makes crunchy dried cheese bars that have taken off as a low-carb snack. By using algorithms to analyze how Just the Cheese’s search ads performed on Amazon’s site, the ad agency Quartile Digital noticed that people who searched for keto snacks and cauliflower pizza crust, both low-carb diet trends, also bought a lot of cheese bars. So Quartile ran display ads across the web targeting Amazon customers who had bought those two specific product categories. Over three months, Amazon showed the ads on websites more than six million times, which resulted in almost 22,000 clicks and more than 4,000 orders.

That 20 percent conversion rate — a sale to one out of five people who clicked on the ads — was “amazing,” Mr. Knijnik said. “That is the kind of powerful granularity for building the target audiences that just Amazon can give you.”

Like other ad networks, Amazon uses cookies and other technical tools to track customers from its site onto other websites. They let the company know that a person who recently bought a diet book is now reading news on CNN and could be targeted on that site with an ad for a protein bar. Amazon does not tell the advertisers who that user is, but it does serve her ads on the brand’s behalf.

“They want more branding dollars, those TV dollars; that is really part of the reason they launched that advertising attribution data,” said Melissa Burdick, who worked on Amazon’s ad teams before leaving to found Pacvue, which provides tools for brands to optimize Amazon ads.

Aside from shopping habits, consumers give Amazon other information that it uses to aim ads. For example, Amazon’s website has a section called “Garage,” where customers can submit their car’s make and model information to make sure they buy parts that fit. In 2015, Amazon used Garage data to help an auto insurer target specific customers, according to a case study on its site. Now, on their own, brands can choose to show ads to drivers who have an Acura MDX as opposed to people who have an Acura TL via Amazon’s ad portal.

Amazon has slowly been developing ways for brands to target their own customers and shoppers, like other ad networks have offered for a while. Advertisers can upload their own customer lists, which Amazon matches with its database, and then show ads to those customers, or other people Amazon’s algorithms determine are similar.

Mr. Denny of CAVU Venture Partners pointed to one of his firm’s investments, One Brands, which sells protein bars on Amazon. The company can retarget customers who looked at the product page for its birthday cake-flavored bars, as well as so-called look-alike audiences that lets One Brands find consumers whose shopping behavior Amazon has determined is similar to people who have bought the bars before.

It not only finds those customers, but Amazon automatically shows different ads to different people based on their shopping behavior. People who do a lot of research on products may see an ad that features positive product reviews, whereas those who have signed up for regular deliveries of other products in the past might see an ad offering a discount for those who “Subscribe & Save.”

“Early tests are showing this is insanely powerful,” Mr. Denny said. “They can do this and nobody else can come close.”