Long before MediaMath shut down, Mike Hauptman and Dan Bougourd were intrigued by the idea of a meta DSP.
For many people it seemed like an allusive fantasy: A single DSP that could interact seamlessly with all DSPs, enabling media buyers to launch and optimize campaigns from a single interface and acquire inventory through whichever platform is best suited to achieve campaign KPIs.
Then came the bankruptcy of MediaMath and the idea of a universal DSP took on new urgency as media buyers were forced to learn new interfaces in a matter of days.
Long before MediaMath shut down, however, Mike Hauptman and Dan Bougourd were intrigued by the idea of a meta DSP. They both joined MediaMath in the company’s early days, but by 2018, they were itching to bring programmatic advertising to the mid-market advertiser.
They pitched the idea of spinning off a new company that would offer a meta DSP built on top of MediaMath to their bosses, who were game. That year they launched AdLib, a DSP designed to bring programmatic advertising to the SMB market through streamlined workflows.
Mike Haupman met with AdMonsters to talk about meta DSPs and the drivers that led them to build and launch one.
AdMonsters: You and Dan are programmatic old-timers. When did you first join MediaMath?
Mike Haupman: I joined MediaMath back in 2010 when the company launched the first self-service DSP. I was the first sales engineer and spent the next seven years building and leading the Global Technology Solutions team. Dan joined a year after me, in a similar role, supporting the expansion into the EMEA region.
Launching AdLib — A Meta DSP
AdMonsters: What made you and Dan decide to launch a new company?
MH: We wanted to make DSP advertising easy. The first tagline of AdLib was, “The Premium DSP for Everyone.” We saw the challenges that media buyers faced trying to access premium ad platforms such as MediaMath and others.
For media buyers at any size agency, it’s a complex business to learn a DSP interface, set up campaigns, and optimize them on an ongoing basis. And doing things like automation and scaling is incredibly resource-intensive. Now increase that complexity by factors when trafficking campaigns across multiple DSPs.
For midsize buyers, those obstacles are formidable. On top of this, they faced additional barriers to entry in the form of commercial commitments and large monthly and annual spend minimums. These challenges combined to effectively bar midsize marketers from participating in programmatic advertising.
AdMonsters: To encourage widespread participation, you started by building a DSP for midsized marketers on top of MediaMath?
MH: Yes, at the time Dan and I were both intimately familiar with MediaMath’s APIs and customizing them. We wanted to explore new options and scratch our entrepreneurial itch, so we approached the MediaMath founders to see how they felt about us launching a really easy-to-use version of MediaMath. That’s what we did. Today, marketers of all sizes use AdLib as a meta DSP to launch and manage campaigns on top DSPs across all channels & screens
AdMonsters: Talk a little more about the concept of a meta DSP. What is it, and is it your goal to put all of the other DSPs out of business?
MH: Our goal is to work with all of the DSPs, leveraging the best of all worlds on behalf of our clients.
The industry has talked about a meta DSP for a long time, which is a single interface that allows the user to login and run media across whichever DSP or DSPs are best suited to meet the object of the campaign.
For instance, if it’s a CTV campaign, Beeswax is a great option thanks to its efficient connections to CTV supply. If it’s a performance-based campaign Criteo is the way to go, because that platform is just incredible at driving performance-based programs across display and video.
A big benefit of a meta DSP lies in the opportunity to have a simple workflow tool. The meta DSP serves as the system of record for users, and they use the simple interface to build and manage campaigns, but AdLib, not the user, determines the best DSP to activate the campaign and manages the connections, gathers the reporting, and displays the consolidated results in a dashboard.
AdMonsters: So media buyers are still using multiple DSPs, but doing so through one interface?
MH: That’s right. Media buyers today access multiple DSPs so they can maximize results and minimize the risks of missing key audiences. But ask any of them and they’ll tell you it’s a time-consuming and error-prone process to set up campaigns in each one.
From an AdOps perspective, adoption, trafficking, and workflows are very different. It takes a lot of investment to teach people multiple systems, and employee churn leads to a loss of those investments.
The other challenge a meta DSP addresses is business continuity if a DSP, like MediaMath, goes dark. This is a reality that a lot of media buyers faced this past June.
Dynamically-Allocated Budgets
AdMonsters: How does AdLib determine the best way to allocate budget for a campaign?
MH: We’re building out a concept we call dynamically allocated budgets. Essentially, we look at the DSPs that are delivering the strongest performance per campaign and focus media spend there.
Let’s say a user sets up a campaign with a CPA goal of $10.00. The system will allocate 50% of the budget to one DSP 50% and the other 50% on another. AdLib will then determine over time which DSP or which set of DSPs delivers the best results for that budget. We further fine-tune the performance with media-mix modeling, as well as dark market / light market testing.
The concept of cross-DSP segmentation is something we’re still testing and developing.
AdMonsters: What is dark market and light market testing?
MH: Let’s say you start with two markets that are relatively similar in terms of size and demographics. Historically, media buyers run media in both markets with similar budgets and benchmarks. In other words, the buyer tests DSP A and market A and DSP B and market B. After a bit, the buyer compares the results to determine which market got the best lift from its baseline, and ultimately, which DSP drives better performance.
A meta DSP improves on this important process in a few ways. First and foremost, it eliminates the potential of user overlap, as the tests occur in different markets. Second, agencies can assess performance without holdout testing, or using pixels to segment users, which gets really messy really fast. Dark market vs. light market testing is cleaner and delivers a stronger signal without the complexity we normally associate with market comparisons.
AdMonsters: How has MediaMath’s bankruptcy affected your business?
MH: Prior to June we primarily worked with midsize media agencies, but since integrating with more DSPs beyond MediaMath, we have begun working with larger agencies. They’ve been receptive to AdLib because they too are struggling with the fragmentation and the soaring labor costs that are part and parcel of using multiple DSPs for their campaigns.
Publishers Get Audience Extension and Scale with AdLib
AdMonsters: Do publishers use your platform?
MH: Yes, publishers use our platform for audience extension services, which has traditionally been a highly labor-intensive task. There’s a lot of manual labor involved in finding their buyer’s audiences on the open web. A meta DSP automates this for publisher AdOps teams.
We’ve worked with several publishers in audience extension services where no humans were involved whatsoever. These campaigns are simply marked for amplification or extension, and they flow the platform and are activated across the appropriate DSP. Think of it as audience-extension-in-a-box.
AdMonsters: This sounds like a great way to scale campaigns without cookies.
MH: We definitely see a future for first-party data activation.
AdMonsters: And speaking of the future, a meta DSP seems like a good way to future-proof the media buyer’s ad ops.
MH: That’s true. While DSPs may come or go, the trafficker needs to worry about just one interface.
About Mike Hauptman – CEO & Founder – AdLib
Mike is a programmatic marketer with over 17 years of experience solving complex and large-scale technical business challenges for Fortune 500 brands, agencies, and advertisers.
Prior to founding AdLib, Mike was one of the first 100 employees at MediaMath, where he held various roles, including VP of Technical Business Development and Global VP of Platform Integrations.