“Data-Driven Thinking" is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Louis Moynihan, vice president of product innovation at Demandbase.
Most ad tech participants don’t ever need to deal with choosing between custom and generic bidders.
A bidder is the piece of the ad tech stack that helps buyers cherry-pick the impressions they want to purchase in real-time auctions. Most companies choose bidders that are readily available off the shelf, rather than building this technical infrastructure from the ground up.
For a small subset of technology companies and newer participants, however, this is a huge and strategic decision that, if left unanswered, may result in a vanilla offering. After a few years into a product evolution, this indecision not only risks low differentiation, it may be too late to start again.
While not everyone is directly impacted, the industry at large should care because there are a number of newer would-be industry participants not interested in renting seats or sharing too much of their proprietary data. If custom bidder advancements attract more vertical entrants and ultimately more advertisers, this only increases the ad tech economy and produces new opportunities.
Building a custom bidder is not the right choice for everyone. The ad agency business, for example, has a reputation for not building technology but renting best-of-breed technology to support what mostly has been a services-oriented business. Ad agencies should not and did not build their own bidders, which why the term “trading desk” was coined to represent the services that sit on top of a white-labeled ad tech offering.
But in other circumstances, building a custom bidder was the only option. Video bidders, for instance, were created because display bidders in 2010 didn’t offer any real video bidding. Not only were custom video bidders created for platforms such as Adap.tv and LiveRail, but larger display demand-side platforms were also forced to build their own video bidding features.
Mobile bidders went through the same type of evolution, but data management for mobile was slightly more complex than for video. Just as video bidders needed to process more complex ad formats, custom mobile bidders also had to rationalize mobile IDs with cookie IDs.
Full-Stack and Custom Bidders
Full ad tech stacks offer a robust and streamlined set of tools for prospective customers who want to rent a seat and transact on their platforms.
Google, AppNexus, MediaMath, Turn and The Trade Desk all have full stacks and feature rich offerings. Of course, you can expose chinks in their armor and slightly differentiate them from each other but, for the purpose of this piece, let’s place them in the category of generic full-stack bidders.
On the other end of the spectrum are custom bidder alternatives that allow new and existing participants to build their bidder from scratch.
New Entrants, Brands Adding More Custom Technology
A prime candidate for building a custom bidder may be a new industry participant sitting on exclusive and sensitive data. The obvious example here is Amazon.
Do you think Jeff Bezos would have allowed Amazon’s data to be fully explored and activated by a Google stack five years ago? No, Jeff Bezos required a custom bidding solution as Amazon dipped its toes in the water, and so would I.
Other brands that publicly talk about their custom bidders include Netflix and Verizon. I expect this trend to continue, especially in specific industries with exclusive data that must be well protected.
There are more obscure examples, such as SAP/Hybris, an ecommerce technology provider. As SAP XM explored getting into the advertising business, there would likely be a very serious discussion around building its own nuanced stack and full ownership of the product road map and resources. Activating SKU-level data from retailers inside a DMP and DSP, while staying mostly inside the SAP firewall, is not only a great value proposition, it’s also a differentiator.
Another candidate for a custom bidder: international entities that would refuse to rent a seat on an American stack due to pure geopolitical viewpoints. However unjustified, this happens.
There are very compelling reasons to consider a custom bidder in these examples, but it is the level of exclusivity that should be the determining factor for diving deeper into a custom bidding evaluation.
Custom Bidder Downsides
There are two huge disadvantages in doing it alone and building a custom bidder from scratch.
First, the time to market is much longer as you have to build many bells and whistles to derive the full benefit. It takes a lot of work to get up and running and maintain; for the organizations that are not adequately staffed, this can be crippling.
The second issue is the server and listening costs, which can run into millions of dollars per year. This can be challenging to fund for early- to mid-stage business models.
The ‘Ad Tech Grid’
The AppNexuses of the world have an advantage in volume of ad calls and standard formats and are offering some custom capability, but they lack full customization. In the not-so-distant future, Google and AppNexus have the opportunity to offer a hybrid bidder. This will have all the advantages of the generic stack coupled with custom capability.
Think of all major pieces of the full stack like an urban electricity grid. In urban planning, dams and waterfalls produce electricity, with the wholesale distribution delegated to states and counties, which then have utility companies handle the maintenance and billing to end consumers. If you build a new house, you tap into a mature and reasonably maintained grid of electricity and water.
What I call an “ad tech grid” would require a separation of the bidding logic from the server and infrastructure costs, rather than all the components bundled together as they are now. In theory, a marketer would pay a vendor like AppNexus for its generic ad tech platform and separately pay Amazon Web Services for its server and infrastructure costs associated with usage. The marketer would also pay ad exchanges and publishers for the media they purchase. These pieces of the supply chain could and should be broken into individual components.
It may also be possible to push more of the real-time bidding decisions to the network edge, closer to the audience, further enhancing a utility grid effect. This would allow new ad tech companies to swap out any one of these pieces and build new custom features, ultimately making ad tech more extensible.
I do see signs that leaders in custom bidding solutions are developing the equivalent of an ad tech grid. It will be commercially difficult to combine tailored listening with all the ad tech features now expected. Therefore, the ad tech grid is not yet a reality today, but I think we are close. Ultimately this would lower entry points for a new flurry of value creators. As an industry, it is something that should be discussed more.
Custom bidder is a loaded term, but for the health of our industry we have to bring new entrants into our ecosystem. We must attract net new value creators, and we will always need new formats and channels that are going programmatic. Custom bidders and an accessible ad tech grid are part of our needed diversification and would add a rich texture to our ad-tech future.
One big question remains: Will an ad tech grid come from the custom bidding vendors or the leading generic stacks?