Q&A: How marketers can keep pace in the DSP features race

In the following Q&A, Malone talks about the challenge automating campaign management brings to the table and unpacks the steps teams can take — beyond simply expanding the number of hands on keyboards. The conversation highlights ways that digital marketers can proactively solve the challenge of ever-expanding DSP feature sets, now and in the future.

What are the major challenges agencies and brands are grappling with in running effective digital campaigns today? 

Brian Malone: The tools for running a successful campaign have evolved over the past decade. What once was a simple campaign with a few lines easily managed by a small agency is really a thing of the past. Tech automation, and specifically DSPs, have evolved faster than our human capability to manage them manually. Automation has effectively created opportunities, while at the same time it has made tremendous challenges in how to run campaigns and get the most ROI for the client. 

Why do you believe the DSP feature set has outpaced a marketer’s ability to use them?

Malone: They know what the features are, but they don’t have the bandwidth to use them. We’re finding that teams must meet certain goals with their in-house staff. But those internal teams are not hitting their goals because there is a gap — they just don’t have the bandwidth, or they have competitive strengths in other areas.

For the marketing team, what are some of the other pitfalls around DSPs offering so many features?

Malone: One of the most important things to think about is that many solutions are very focused on increasing or decreasing a price that the organization is actually willing to send out into the marketplace. So, for example, if a particular inventory isn’t very good at reaching the team’s goal, they’re going to bid-factor that inventory down by a little bit or completely exclude it. That works most of the time, but sometimes the marketer is just paying more or actually overweighting certain inventory sources. There are things that we’ve had to do within the DSPs that we work with, such as creating a more diversified spread and trying to reach the target market and not concentrating so much on certain audiences.

In all this attention to the details, what do you often see slip through the cracks?

Malone: This is really important: If you have a proliferation of different tactics to drive performance or there’s just so much more granularity to understand, there’s often very little day-to-day updating of the investment across different tactics to maximize the goal. It could be as simple as looking at video versus display to understand which one is doing better on a CPA basis. Let’s allocate more to video if that’s the case. But again, it’s almost like it’s so overwhelming just to make sure the tactics are on pace that you can’t take advantage of all these other great features.

What are some examples of features that marketers likely don’t use today?

Malone: Audience and inventory optimization are great examples. Marketers have an idea that they’ll execute time-of-day type optimizations. Yes, that’s one part of it, but time-of-day with a certain user type is probably the better fit. So, you have all these multivariate correlations that go on around optimizations that may go underused or not be used at all across teams.

What should marketers consider when looking to use more of these features?

Malone: Step number one is determining whether marketers have access to resources that could take advantage of those APIs. They need to be able to take advantage of the algorithms and APIs in these platforms. Once they get into a situation where they have access to these features, they would need to understand the cost-benefit analysis of using them. For this, a good experimental framework is required. We see this a lot with our stronger brands, where there’s an effect in which the algorithm will start looking for cheaper and cheaper inventory. And then that inventory keeps going down until it gets to this natural conversion rate. But that’s your natural CPA goal — you’re not actually adding any new people into that market share; rather, you’re just finding the natural floor. What you want to do is show incremental lift from what would be naturally occurring and then find how many more conversions can you get above and beyond that, and that should be your true CPA.

How has your organization changed to adapt to this feature race?

Malone: In the past two decades, we had traditional campaign managers and more ad-ops personnel. Today, we have more analysts, engineers and data scientists — people who can work at scale but who don’t necessarily know the UIs of the DSPs that we use. But that’s OK because they are actually better at managing campaigns than maybe the traditional campaign managers because they can do more with the hours in the day. 

The team dynamic is really focused on automation, but there is a caveat — with automation, you need discipline and structure on how to actually set up your campaigns. There are details that are easy for humans to recognize but harder to implement. 

Our team at Nexstar Inc. is focused on analyzing the data and interpreting what the tool is helping them monetize against. This helps them make informed decisions on optimizations in real time, and then they use automation to execute on their strategic guidance at scale. 

The team also works in different verticals based on specific goals and KPIs to craft solutions and then use automation to implement it on a wider scale than with manual management of the campaign. What we’ve found works in QSR may have a unique set of variables and audience traits vs. managing campaigns in financial services or travel. 

At the end of the day, it’s about balance — automation balanced with an expert team and structure so that we are able to do these things at scale.

What is the number-one thing that a digital marketer can do to keep up?

Malone: Whether you have the very best automation or you’re doing everything manually, the central theme is that you have strong KPIs and outcomes. And if you can consolidate what you’re trying to achieve down to one KPI, and the DSP has access to the data it needs to leverage it, then you get a positive feedback loop. This will give you much more success with whatever features might be built now or in the future because they are typically built around the idea that a dependent variable can be optimized. Algorithms are set up to do all that. If you can just focus on the piece that you’re trying to optimize and make sure that it’s a real-life goal that matters to your business, then everything else will be a lot easier.