As a marketer in the mobile space, you have likely heard the word “programmatic” thrown around quite a bit. In 2021, 88% of digital display marketing in the U.S. is projected to be done via programmatic advertising — two-thirds of which will be done on mobile. With that said, it is safe to say that including programmatic inventory in your overall marketing media mix can be incredibly impactful from both a reach and diversification standpoint. If you are looking for a better understanding of what exactly programmatic advertising is — or even if you are already familiar with it but are working on scaling or increasing your existing programmatic channels, then you are in the right place.
Taking a step back — What is programmatic advertising?
The high-level definition of programmatic advertising is the purchasing of ad inventory using real-time bidding (within a matter of milliseconds). Programmatic inventory is inclusive of in-app, mobile web, desktop — even CTV and podcasts.
If you are completely new to the concept of programmatic advertising, you can think of the buying and selling of programmatic ad space like the buying and selling of stocks on the stock market. The buyer (you) sets a price they would like to purchase the ad space for based on market variables. The seller (exchanges) accepts the highest bid and continues to sell off more at lower and lower bids as the stock decreases in value (think of this like ad placements throughout a user session).
Develop your strategy — What processes and partnerships do you need to have in place in order to run a successful DSP channel?
Identifying your KPIs and establishing a strategy are two incredibly important steps before launching any programmatic campaigns. For example, if your primary focus is reach, then you may want to focus on cheaper ad placements and a wide inventory scope in order to serve as many impressions as cost-efficiently as possible. If your KPI is for a more down funnel action, you may want to start with a more refined inventory approach and higher impact ad placements. Since there is a wide variety of programmatic inventory to buy, it is important to be strategic in what sizes you choose to run from both creative resourcing and KPI standpoints. Banner placements, for example, account for over 12% of the overall mobile programmatic inventory and typically have some of the cheapest CPMs. Higher CPM placements such as full-screen video, interstitial, and MRAID creatives are all additional ad units you can tap into o find what media works best for your marketing strategy.
Data sharing pre-campaign launch is also important in order to give your campaigns the most efficient possible start. If you are able to share a suppression list of existing users with your partner, that is a solid foundation in order to not waste your budget on existing users. On the flip side, utilizing programmatic for retargeting will also require a list of user identifiers in order to launch the campaigns. Outside of device lists, you can also pre-filter the inventory you are buying based on features such as IAB category and device-level features such as OS version and connection type. You can also share historical purchaser or event data with your partner in order to ramp up the models prior to launch. Outside of first-party data, third-party data can be used in order to layer on additional targeting — or as a starting point if you have a clear demographic you would like to reach or suppress but lack the first-party data required to do so. Lastly, you should leverage your partner if you are using a managed service to provide you with insights and best practices for a campaign launch.
Choosing your partner — What type of DSP channel ownership is best for your team?
Last, but certainly not less important, is choosing how to operate your DSP channel. I have listed some of the primary options companies take in their approach to running a DSP channel in the chart below, with the X-axis being the level of managed services and the Y-axis being media fees.
- Fully In-House DSP — Building a fully in-house DSP can be the most cost-effective option and allows you full data transparency without having to share any of your precious first-party data with an external partner, but has many internal requirements that need to be met in order to be successful. This approach takes a significant amount of more in-house support, varying from media buyers to data science and engineering in order to build and maintain the bidder as well as the underlying data. Having a fully in-house DSP can also be expensive purely from a data-gathering standpoint as you ramp up the learning process for your models from scratch.
- BAAS — Using a bidder as a service allows you to develop customized models without needing to actually build out your own bidder in-house. This solution can be incredibly powerful in many of the same ways a fully in-house DSP is. You are able to customize your own models and directly access data all of the way down to the bid request level. Some BAAS partners also have the ability to run across many different types of programmatic inventory, so this is important to take note of if you are planning to run on other programmatic inventory such as CTV or desktop in order to manage all of your programmatic buying in one space. Cons for using a BAAS are similar to the ones for running a fully in-house DSP in terms of needing in-house personnel resources and time to build and develop your models. Additionally, using a BAAS will incur platform fees and potential additional fees for items such as bid log access.
- Managed Partner with Self Service Access — This is a good solution if you have the resources to handle more direct campaign management but do not have the resources to build out or maintain your own bidder and models. Using a self-service option can also allow you more granular access to campaign and inventory data. Downsides here are that you will typically be paying fees on the inventory you are buying, although they can come in lower than fully managed services at a 10-15% range and may decrease as you scale.
- Agency — Agencies are great for teams with limited resources because they can handle everything from campaign management to creative depending on the partner. Of course, with these fully managed services come high fees that are not always transparent and can be upwards of 25% of overall media spend. Furthermore, your access to more granular campaign performance data may be limited.
To summarize, selecting the best option for running your DSP channel consists of many factors to take into consideration. Choosing the best option for your company requires analysis of both your internal resources as well as the scale you plan on running at.
Launching your DSP channel — What should you look out for after the initial campaign launch?
Something important to keep in mind when launching your DSP channel is that your DSP needs data in order to develop its models and optimize the bidding strategy. This is only exacerbated if you are sharing limited data with your partner or are optimizing towards a down funnel event with a high cost per action. While the models ramp up, it is especially important to keep an eye on the publisher/exchange level performance as well as device metrics as these are fields you can manually filter or increase spend on outside of solely relying on the models to do it for you. These features are also useful in identifying potential fraud by looking at metrics such as CTI and retention comparisons across publishers/channels for abnormalities.
Growing your channel — How do you scale efficiently?
Whether you are trying to maintain the success of your campaign as you scale or need to reevaluate the campaign structure due to poor performance, the beauty of programmatic advertising is that you as the marketer you have much more control and transparency in regards to campaign performance (compared to other “walled garden” channels). Here are the three main focus areas for campaign scale and optimization:
- bid models
From a creative perspective, launching in additional sizes is a great way to scale and impact your overall performance based on your goals. You or your DSP partner should be able to see the number of requests coming through on the inventory you are running for various ad sizes. Typically, banner placements account for the most amount of bid requests as they make up roughly over 12% of overall mobile ad unit inventory. This can also be approached by specifically opening up additional ad placements on your top-performing inventory. It is important to note that your CPMs will fluctuate depending on the placement sizes you decide to run, typically increasing with placement size. The incrementality of ad units will also vary, so it is important to run ad placement incrementality tests to ensure the placements you are running are truly impactful for your business goals.
When it comes to inventory, launching on additional ad exchanges is a great way to scale or test new inventory for performance efficiency. You can slowly begin to test additional exchanges based on partner recommendations, or you can take a more data-driven approach (given you have enough data to work with) and assess what additional exchanges and publishers to add into your campaign targeting based on where your top converting users are coming from. If you already have a list of top publishers, you can also set up a PMP deal in order to secure more volume as well as have more control over where your placements display within a user session.
Lastly, you can test different models and optimization tactics in order to find what works best for your product. For example, if your primary KPI is a purchase event, you can test optimizing on earlier parts of the user funnel. An example of this on a mobile gaming product would be using tutorial complete or a certain checkpoint/level event that signifies a rough amount of hours in-game. Whether you are developing your models in-house or using models developed by a partner, I highly recommend you test different optimization events in addition to model variations in order to find what works best for your product.
Maintaining your channel — How do you keep your campaigns successful long term?
Leveraging your first-party data is important across all stages of the campaign management process. Keeping a tab on not only your high-level campaign performance but more granular data points from device-specific — such as OS version and device manufacturer — all of the way to inventory-specific metrics — such as publisher and exchange — can be incredibly powerful in determining where you should (and shouldn’t) focus your efforts. This is only compounded if you have access to your programmatic bid requests as those open up a wide variety of potential uses, such as developing high-valued user lookalike profiles. If you are running with a managed partner and do not have direct access to your bid logs, you can either ask your partner if they are able to share some of the data with you or discuss the ways in which they are using that data (outside of model development) to enhance your campaign performance.
In addition to using first-party data, using external tools such as App Annie or Sensor Tower (this isn’t a paid promotion) can be incredibly powerful in gaining insights on the creative landscape for your product, viewing competitor metrics, and even finding new publishers to target through cross-app usage analysis. If you are marketing an app and are not regularly checking tools such as these, I highly recommend you do so for even non-programmatic efforts.
Lastly, it is important to keep an eye on how the programmatic landscape shifts over time — exchange and publisher performance fluctuates and the landscape has become even more complex with the release of iOS 14.5. Focus on the foundational aspects of mobile marketing campaigns that you have control over (creative, inventory, and model methodology) and work your way up from there.
Alexa Wieczorek is a User Acquisition Manager at Electronic Arts, specializing in programmatic advertising. Her background is in the mobile gaming space, having previously worked at DraftKings and Glu Mobile.