There is a lot of buzz in the digital advertising industry about “Programmatic Marketing.” Some people describe it as the intersection of big data and technology. Others primarily think of it in the context of Real-Time Bidding (RTB) on display exchanges. I thought I would take a stab at describing Programmatic Marketing and RTB for people who are interested in these topics, but are not very familiar with them.

What exactly is Programmatic Marketing, and how does it improve ROI?

At a high level, Programmatic Marketing is the practice of implementing an automated set of business rules to efficiently target your most valuable customers and prospects with personalized ads.

Each of the bolded phrases in the statement above speaks to the promise of programmatic marketing from an ROI standpoint.

  • Automated. By automating buying decisions, marketers remove the friction of the sales process (including humans placing buying orders) and reduce their marketing costs.
  • Efficiently target. With programmatic marketing, the goal is to eliminate wasted impressions and clicks. Thus, you only show ads to users who have intent, and who are likely to take the desired action (e.g. “buy”).
  • Personalized ads. Programmatic marketing increases the likelihood of consumer action by showing each user a personalized message. The goal is to present users with a more customized call-to-action based on their recent browsing behavior, for example, or other anonymized data that you know about them.

There are many forms of programmatic marketing. In the display world, programmatic marketing often takes the form of dynamic retargeting on Real-Time Bidding (RTB) exchanges.

Why is RTB important?

RTB is one of the fastest growing parts of digital marketing, and it holds significant promise to increase marketer’s ROI. According to eMarketer, RTB Digital Display ad spend will grow from a $1.9B market in the US in 2012, to a $8.5B market by 2017, a growth rate of 35% CAGR. As a percentage of total digital display ad spending, eMarketer predicts RTB will grow from 13% in 2012 to 29% by 2017.

US RTB market forecast (source: eMarketer)

Although RTB is projected to grow enormously over the next few years, both publishers and media buyers have some concerns with it. For publishers, the concerns are primarily about maintaining pricing control and direct relationships with advertisers, as well as about the quality of ads that are showing up to their users. As a result of these concerns, publishers may only open up a portion of their inventory for Programmatic Marketing and RTB. For example, some publishers experiment primarily with opening up “remnant” (unsold) inventory to RTB. Additional concerns from publishers include data leakage (DSPs are able to see impression requests for users without placing bids, allowing them to refine their own segmentation) and potential channel conflict with direct sales.

On the media buyer side, the concerns are around being able to access more premium inventory, and having exclusive or preferred access to inventory. Marketers may select to work with a Demand-Side Platform (DSP), for example, based on the amount and type of inventory it has access to.

How exactly does RTB work?

The basic operation of RTB is as follows. A publisher will have software from a Supply-Side Platform (SSP) to manage and maximize the value of its ad inventory. A media buyer will have software from a Demand-Side Platform (DSP) to manage its marketing campaigns across a number of different publishers and SSPs.

When a user visits a publisher’s website and requests a web page in her browser, the SSP will attempt to fill the ad inventory created by that user’s request. The SSP will make a request to the RTB exchange on behalf of this user, potentially annotating the request with data on the user or the content being browsed on the publisher’s website. The RTB exchange will then fan out this request to multiple DSPs, each of whom may have a cookie on this user’s browser.

Based on its knowledge about this user (e.g. the user recently searched for flights to Hawaii on a travel website), a DSP will bid on the right to serve an ad to this user. The RTB exchange will then run an auction for the ad impression generated by this user. The winning DSP will serve a creative — potentially a dynamic display ad with personalized content, perhaps including the recently browsed flight details, price, and image of the destination — to the user.

As a final step, the publisher’s web page loads in the user’s browser, including the personalized ad that was served by the DSP. This entire process—from user requesting the web page, to the execution of the ad auction, to the personalized ad inserted into the web page—happens in real-time, on the order of milliseconds.

Framework for Analyzing RTB: Targeting, Creative, Measurement

For digital marketers, the goal for every marketing campaign is to reach the right audience at the right time, deliver a compelling message, and then optimize based on the results.

This goal can be decomposed into the framework of Targeting, Creative, and Measurement:

  • Targeting: “reach the right audience at the right time”
  • Creative: “deliver a compelling message”
  • Measurement: “optimize based on the results”

Let’s examine dynamic search retargeting on RTB display exchanges in the framework of targeting, creative, and measurement. For this analysis, let’s consider a theoretical example that involves a user — Rebecca — and a marketer, Costco Travel. Rebecca is researching vacations in Hawaii, and Costco Travel wants to connect with users who have intent to travel to Hawaii. (Disclaimer: for simplicity, I have created a fictitious example involving Hotels.com and Costco Travel. I don’t know if Hotels.com or Costco Travel participates in RTB dynamic retargeting or not, but just chose these brands in my theoretical example.)

Targeting

The big shift in Targeting for RTB is that it’s individual user-driven, not segment/content-driven. With traditional digital ads, a media buyer would buy ads based on the content of a publisher’s website, as a proxy to reach a desired audience. For example, in the past, a travel marketer would purchase ad inventory on the travel section of a portal’s website to promote vacation packages to Hawaii.

In the world of RTB, marketers can precisely target individual users whom they believe have demonstrated purchase intent, rather than the coarse-grained approach of targeting an entire audience. Since the RTB exchange runs an auction for each impression, marketers can set business rules to only bid on user segments that are important to them. And marketers will be able to identify those users who are important to them because they (or their DSP partner) have a cookie on the user’s browser, which also has associated first-party or third-party data.

In our example, Costco Travel is offering a special vacation package to Hawaii, and can choose to only serve ads to users who have searched for accommodations in Honolulu on Hotels.com. They don’t need to waste their impressions on users who visited the Travel section of NYTimes.com, but who aren’t interested in a Hawaiian vacation. Costco Travel’s DSP has the ability to place a retargeting pixel on the Hotels.com website.

Rebecca visits Hotels.com and searches for hotels in Honolulu. During her visit, Rebecca’s browser receives a cookie from Costco Travel’s DSP that includes the data about her search (e.g. destination, desired travel dates, filters used, etc.).

DSP places a cookie on user’s browser

Suppose Rebecca doesn’t book her travel during this session. Instead, she decides that she wants to do more research later, and instead switches gears and visits a news website to catch up on the day’s news. The news website works with an SSP, and the ad inventory that is created by Rebecca’s visit is eligible for RTB bidding.

When Rebecca shows up in the RTB auction, the DSP that represents Costco Travel will bid higher for the right to show her an ad than a random web user, because Rebecca is “in-market” for a Hawaiian vacation. And the DSP will bid lower — or perhaps not even bid at all — for the random web user who has not shown any intent to travel to Hawaii. Thus, Costco Travel is able to direct its marketing budget towards only those users who are in-market for a Hawaiian vacation, and therefore avoid wasting impressions on users who have no desire to travel to Hawaii.

Creative

In addition to only targeting users who have expressed recent intent, RTB retargeting involves delivering personalized ads to user. In this case, the DSP that represents Costco Travel has fairly precise data about Rebecca’s actions on Hotels.com. For example, it may know the destination, number of nights, price ranges, and filters that Rebecca used in her search, which gives them a sense of what vacation packages and pricing would probably be attractive to her.

DSP knows the destination, price range, and other data from the user visit

With this information, the DSP representing Costco Travel can assemble a personalized ad creative that will be most effective for Rebecca. Since they know what price range is likely to be effective for her, and the star-rating of the accommodation she has browsed, the DSP presents a customized ad just for her.

Personalized ad with pricing, destination information

Measurement

RTB retargeting uses the same measurement approaches that direct response (DR) marketers employ in all other situations— it looks at click-throughs (CTR) and conversion rates. Since RTB retargeting may also involve serving a personalized ad to a user multiple times before that user converts, some marketers use view-through attribution as one method of analyzing the ad’s performance.

With view-through attribution, the marketer will establish a window of time — the “attribution window” — during which the user is in the midst of a purchase cycle. For a large vacation, the attribution window used may be a week or two. For a car purchase, the attribution window may be a couple of months.

The “view-through” part of attribution refers to the idea that an ad should receive at least some credit for driving the conversion just for being seen, even if there was no user interaction with the ad. Proponents of view-through argue that data shows that increased frequency of ad exposure correlates with higher conversion rates from users. Detractors of view-through argue that without any user interaction with the ad, you can’t prove that a user actually saw it. The truth is probably somewhere in the middle.

Given RTB retargeting’s extensive use by Direct Response advertisers, the ability to measure conversion performance (including site visitation) is of critical importance. Measuring conversion performance, however, is one of the challenges for RTB retargeting on mobile.

RTB Re-imagined for Mobile and Native Advertising

RTB retargeting is primarily being pursued with traditional display advertising, and is still relatively early with mobile advertising. I have written previously about how mobile ROI is difficult today due to consumer shopping behavior on smartphones and the fragmentation of Internet usage across devices. The fact that users tend not to convert directly from smartphones makes RTB retargeting a challenging proposition on mobile. The way that RTB retargeting will initially develop on mobile will be via driving users to perform in-app conversions, on apps that they have already downloaded to their smartphone. However, in order to tap into the full RTB retargeting opportunity, the industry will have to solve the problem of targeting and measurement across devices.

Similarly, RTB retargeting is also in the beginning stages with respect to “native” advertising. One of the hallmarks of native advertising is that the ads blend seamlessly into the consumer experience, and often strongly resemble content. The ads used in RTB retargeting with traditional display are often little more than a call-to-action to buy, with product information, pricing, and a product image. These ads probably look more like ads than content. In the context of native advertising, RTB retargeting ads may therefore not be able to blend in as easily to the consumer experience. Thus, RTB retargeting ads may need to be re-imagined for native ad platforms.


Programmatic Marketing and RTB are important concepts for digital advertising, given their projected growth rates and the improvements they create for marketer ROI. For media buyers, RTB retargeting — one of the most prevalent forms of Programmatic Marketing — involves precisely targeting users who have active purchase intent, eliminating wasted impressions and clicks. The particular set of users who have intent consist of users who have recently browsed products on your website (“site retargeting”), or who have recently searched for products on your own or a third-party website (“search retargeting”). A Demand-Side Platform (DSP) will bid on desired users that have intent, and a real-time auction is run for the right to show an ad to those users. If the DSP wins, it will deliver a personalized ad to a user, and the marketer will evaluate the impact of the campaign by measuring click-through and conversion performance. There are some interesting nuances for RTB with mobile and native ad platforms. These are early and exciting days for programmatic and RTB within the mobile ecosystem. I’m looking forward to all of the innovation that’s yet to come.