Basics of Programmatic Media Buying

It’s a long, complicated phrase with a seemingly even more complicated meaning. However, programmatic media buying is also an important concept for media planners and buyers (and even students seeking a career in advertising in general) to understand. Programmatic buying is becoming a buzzword in the advertising industry, and while some controversies exist over its usefulness (that’s a different blog for a different day), there is no denying that buying online display ad space programmatically is starting a lot of conversations in the advertising world.


 But what exactly is it? A recent Millward-Brown study said that advertisers don’t even seem to know—60% of respondents said they “didn’t feel that the ad industry had an accurate and unified definition for the term.” So even though I don’t have direct experience with running a programmatic media buy*, I believe the concept is so important and relevant to advertising trends that it needs to be discussed and defined.

At its core, programmatic buying is simply a way to buy display ads online. 

Historically, advertisers had to manually purchase online display ad space by reaching out to specific websites and buying certain sizes of ads.

Source: callum.stephenson.710

Programmatic buying is a way to make this process more efficient and specific. 

A company that sells cooking products can certainly buy display ads on recipe sites. But with programmatic buying, the cooking products company can use data-driven insights of their typical consumer (say, a thirty-year old woman who also likes to do DIY crafts and has young children), and buy sites that have visitors that are “lookalikes” of this consumer.

Sites are able to tell who a consumer “is” by their previous search and browsing history, obtained through cookies and pixel tags.

Cookies are bits of code that are saved on a web browser and retrievable for other uses later. The cookies have a unique ID that is triggered on various sites, proving the users to be the same person.

Pixel tags refer to a small, invisible GIF pixel embedded in a page (such as the “thank you for purchasing” page on Amazon and other online commerce sites). These pixels can “read” the cookies on a site, as well, and discover further data about consumers by tracking the other sites they visit.

A computer and an algorithm execute programmatic buys. 

However, humans are still much needed in this process to help optimize the ads' targeting. Sometimes the buy will utilize real-time buying (RTB), which is an auction-like environment for buying impressions of display ads online. RTB methods only buy domains, but a programmatic buy can include specific pages on those domains (the weather section of a newspaper website, rather than the newspaper site as a whole).

Programmatic buys can also occur when the advertiser buys a certain amount of digital impressions from a website publisher beforehand.

Advertisers can also retarget their consumers. For example, a consumer who has recently visited a site, looked at a specific pair of shoes, but then did not buy ultimately the shoes can be "followed" around the web. This consumer will be served an ad featuring that specific pair of shoes. (These types of ads often show up on Facebook. You know, the creepy ones that make you swear someone was watching your every move?)

Advertisers are still able to set certain parameters when buying programmatically, as they would normally when considering how they will buy display ads. They can set the bid price, total budget spend, campaign flight dates, and similar specifics.


Having at least a general understanding of programmatic buying is beneficial. From here, we can see where programmatic is headed and the issues surrounding it.




*Full disclosure: I briefly shadowed employees at Amnet in New York City during the summer of 2014 to further understand and ignite my interest in programmatic media buying.