Programmatic Advertising: What to Know Before You Get Started

Lisa Hutt - December 9, 2015

There’s always a “Next Big Thing” and this year it’s programmatic advertising. Everyone’s talking about its promise, but for many of us it can seem like one big black box fraught with intimidating acronyms like RTB, DSP, SSP and DMP. Just how does it work and what do you need to know before you consider using it. (If you’re not sure what we’re referring to here, check out our post that outlines what programmatic advertising is.)

Programmatic Advertising Basics

At the center of programmatic advertising (which I’ll call “PA” to keep things simple) are algorithms that in effect say “serve this type of ad when this type of user is visiting a site at this time of day.” The decision as to what ad to show happens in microseconds as someone accesses a web page.

PA is supposedly not only more efficient for advertisers but also for publishers, helping them to more effectively sell inventory. It involves automated real time bidding (RTB) layered with behavior or audience data, including data you can provide from your own data sets such as email lists.

You can think of PA as a very efficient matchmaker between publishers and advertisers, but a matchmaker that is more like eHarmony and less like Match.com. You have to trust the PA platform knows the best matches for you and your advertising goals and that it will get smarter as it learns more about the type of individual likely to “convert” and when and where that person will be online.

Trusting machine learning to run your campaign efficiently may sound like a challenge for PA first-timers, but remember, you are likely already doing this if you run any search or social campaigns that are automatically being optimized for conversion rate or CPA.

The Programmatic Ecosystem

There are many entities that make up the programmatic ecosystem and the benefit of running your ads with a platform such as RocketFuel or MediaMath’s TerminalOne (two of the platforms we use at MoreVisibility) is that you don’t have to make the connections between the various players:

• Ad exchanges: These are open markets for publishers with inventory, to connect with advertisers who wish to buy space. An example of an ad exchange is Google’s own DoubleClick Ad Exchange.
• Data Management Platforms (DMPs): These are used to collect and analyze data from a variety of sources with the end goal of consolidating user data into a centralized platform. This data is used by both advertisers and publishers.
• Supply-Side Platforms (SSPs): Platforms that allow publishers to enter the ad exchange and make their inventory available at optimal selling prices.
• Demand Side Platforms (DSPs): Platforms that work in concert with ad exchanges and SSPs. A DSP will in effect submit a “bid” to the SSP. The SSP will pick the winning bid and display the ad.

Considerations before Jumping into Programmatic Advertising

Acronyms and complex systems aside, here a few things you should consider before signing a deal with a PA platform:

1. PA is not the Holy Grail. You still have to do a lot of the preparatory work you would do with any other campaign—setting goals, identifying segments and creating pertinent messaging.
2. Be prepared to invest money in display. If you have not used anything more than the Google Display Network for display efforts, using a PA platform will seem to be a huge leap in sophistication; but it also has a much higher entry cost. Programmatic often requires a minimum spend of $10K per month. A certain amount of interaction with the ads has to happen for the system to “learn,” and small budgets simply don’t provide enough volume.
3. You should expect CPM (cost per thousand impression) pricing and become comfortable looking at a performance metric called response per 1000 impressions. Since you are purchasing based on impressions, you will want to have a better understanding of what all of those impressions are getting you and response/1K will help you to determine that.
4. You should be comfortable with the value of “view-through” conversions, i.e., conversions that happen after someone “sees” the ad—because the platform will count these as “wins.” View-through conversions (also known as post-view conversions) have always been a somewhat fuzzy metric for obvious reasons—how can you ever be sure someone actually saw your ad, let alone acted upon it.Keep in mind this trust in “views” has been a mainstay of television advertising for 60+ years. Instead of worrying about the direct connection between the click and the conversion, start to look at the correlation between advertising efforts and a lift in direct traffic or traffic coming in via branded search queries.  We have discovered that it’s best to use a shorter conversion window of 1-3 days for post-view conversions and leave longer conversion windows for post click conversions.
5. Be prepared to focus on the total customer journey. If you are only focused on conversions that are sales or leads (so called “performance advertising”) you may feel PA doesn’t deliver. The platform needs to be allowed to see “clicks” that drive traffic to your website as “conversions” as well. That is the only way it can learn what type of person is interested in your message.
   Just like other display advertising tactics, PA is often more about branding and awareness and less about the final leg of the customer journey. Consider using programmatic for both prospecting and remarketing—using the prospecting campaigns to fill the funnel that you will then be able to remarket to.
6. Make sure you can track the full customer journey in your analytics platform. We can’t stress this enough. The platform will consider a click to your site a “win”— which it is, but you need to be able to translate what that click did for you farther down the path.
7. PA platforms have varying options for campaign management and various levels of transparency. Find out beforehand what type of information you will get during and after a campaign. Ask to see representative reports. We have found that actual placement information is hard to come by, and the most granular info we were able to see was the ad exchange.
8. Not all of the platforms are ready to deal with display advertising newbies and some tend speak as if all advertisers have been doing large scale display buys for years. Don’t be intimidated.

If you do your homework and understand what to expect, PA can be an invaluable tool to reach people you otherwise would have no way of getting in front of and doing it with the efficiency that only machine learning can offer.