Commentary

A Chat With AppNexus Chief Economist Gabriel Weintraub

AppNexus last week announced that it had hired a chief economist, Gabriel Weintraub, to help the company improve its RTB offering. Weintraub actually officially started in the newly created role in March.

Chief economist is a hefty title, one that ad tech/martech stakeholders don’t see frequently. It’s widely known that AppNexus plans to go public, so perhaps having an economist on staff will be helpful. Weintraub is a professor at Stanford University’s business school, and once studied with Susan Athey, the economist behind Microsoft Bing’s auction design.

RTBlog decided to find out more about the complex issues Weintraub will tackle in his role.  

RTBlog: What are the inherent flaws of RTB?

Weintraub: At a high level, the industry has advanced so quickly that a lot of solutions came out of necessity but it’s not clear that these things are the right solutions for the marketplace. Over time, we’ll start to get closer to the best solutions.

Take, for example, low bid density and the second highest bid. We’re sometimes targeting so finely that there’s only one buyer that really wants that customer. Another reason is that the DSP [demand-side platform] doesn’t submit all the bids to the auction, only the highest bids. The question is, if I’m a seller, what’s the best way of auctioning with this phenomenon? Different companies have come up with different solutions. How do you monetize your inventory when you have low bid density?

There’s some research that shows when you think about bid withholding, that it incentivizes sellers to set more aggressive floors. And that everyone, not just the seller, would be better off if DSPs wouldn’t withhold all the bids. If we could come out with an arrangement where DSPs show all the bids, I think both buyers and sellers would be better off.

RTBlog: What about header bidding?

Weintraub: It’s a very important solution to penetrate what used to be a closed market—to open it up to other demand sources. But there are questions. What should the auction rules be for header bidding? Should they be the same as they are for RTB? Or how do you decide on an auction that competes head to head with Google’s AdX?

There’s a tradition in economics around market design and what the rules and logic should be for auctions. And there are rules for different marketplaces. What should the auction logic be for monetizing ad inventory?

RTBlog: What is market design?

Weintraub: It’s an area of economics that uses tools from game theory, econometrics, and micro-economic theory to understand how marketplaces work. It looks at what the dynamics are, and advertising auction logic that affects incentives to achieve certain objectives. It looks at how to set up rules to make the auction more efficient.

If your goal is increasing a seller’s revenue, that’s a goal market design can address. Another goal is efficiency and making markets safe. And you want to monetize the inventory, and outcomes that are fair for both buyers and sellers. Market design is having a significant impact on how digital marketplaces are being run.

RTBlog:What do you think needs to be fixed? And how would you go about fixing it?

Weintraub: It would be helpful to come up with an auction logic that’s transparent for buyers but also effective in monetizing publishers’ inventory, given low bid density.

A/B testing is helpful for many things, but those tests can be a bit deceptive when thinking about auction logic. If you do an A/B test, I want to see how buyers will change their behavior.

If I want to measure the response from buyers, there are other tools I need to use. What is the rational response from buyers? To get an answer to that question, the typical A/B test is not enough. You need to do more sophisticated testing. A/B testing doesn’t pick up how buyers change their behavior in response to changes in auction logic.

Next story loading loading..