Thought Leaders in Big Data: Interview with Andy Nibley, CEO of Yieldex (Part 2)

Sramana Mitra: The situation you described – three or four salespeople targeting the same customer and selling four ads – based on your technology, how does the publisher organize itself to be able to tackle that?

Andy Nibley: Our technology keeps track of targets and the overlaps between inventories. All of those salesmen are pulling inventory from the same pile, but they are not decrementing each one when they take inventory out. That is what we keep track of. When somebody sells to 18- to 24-year- olds, you are going to have to take that out of the female category, the car category, and the region – in this case the New York area. You get a hyper accurate look at your inventory instead of a guess based on sampling.

SM: Essentially what is happening here is that you are able to forecast, so the publisher is going to be able to place and not over promise.

AN: There are three things we do. The first is: How many ad impressions do you have to sell? That is called capacity, and most publishers actually don’t know how many they have to sell. For example, they don’t know how many 18- to 24-year-olds they have to sell. Second, how many have they sold already and how many do they have left to sell? That is called availability. That is really important, because publishers are taking what are called “haircuts.” Because they are guessing at how many impressions they have to sell. A typical example is when a sales manager calls an operation and says, “I want to sell to 18- to 24-year-olds. How many impressions do I have this month?” But they don’t want to upset the advertisers, so it is better to underestimate. If they look at it, they may see that they have two million ads to sell, and they would tell the sales manager that they have 1.5 million in order to not get in trouble if they underestimate. The sales manager then goes through the same process with his sales execs when they say, “Better not take a chance of upsetting advertisers. I am going to tell my sales execs they have one million left to sell.” Right there you take in a 50% cut, and that money is being left on the table.

We see publishers leaving anywhere from 20% to 70% or 80% on the table because they don’t have accurate forecasting. What we do is help them realize that revenue by giving this more accurate forecasting. This you can only do if you look at every ad impression. It is also an expensive proposition, and the reason why you have to be a big data analytics firm to be able to take in all that data, crunch it, and come up with an accurate forecasting. You also need to have what I call “special sauce.” You have to be able to accurately forecast what the ad server is going to serve. That is based on understanding ad server technology, seasonal models, and spike detection models, all of which come with our product.

SM: This was very helpful. Let’s do a couple of more use cases.

AN: First of all, I think if you can stop under delivery, which we have done for almost all of our clients, that is money that you can take off the table. That is money the publishers can make that they weren’t realizing before. Another case study would be from a big e-commerce firm. We help them create new products. You can do this by understanding the overlaps in your inventory. Maybe you are selling entertainment aside from the CPM. If you pulled out movies and off-Broadway plays, you can find that these segments actually command much higher CPMs – $15 to $20 CPMs. If you create those new products, you could be getting more money out of the inventory you have now, and the rest of the inventory you can either sell or try to create even more products out of it.

This segment is part 2 in the series : Thought Leaders in Big Data: Interview with Andy Nibley, CEO of Yieldex
1 2 3 4 5 6