Seeing if you are on the supply side, you need to figure out what price you want to set for your inventory. The simplest way to do this is to put in what's called a CPM floor. That's where each publisher will set a minimum floor price for their inventory. If it is, if an advertiser bids below that floor, then the bid responsive rejected the ad does not fill anything at or above that floor is accepted. You can have different floors for different ad units, maybe the homepage is a higher price and then somewhere a few pages in your video should be priced higher than your banner ads. A 300 by 250 may be priced higher than a 320 by 50.

So there are a lot of different things to take into consideration when putting a floor price in. One thing to keep in mind is that for a waterfall, there can be multiple floors for a single ad unit, one at each level. The first look will get made Be $1 floor. But if they can't fill and we go to the next partner there, you got to be at a lower rate, maybe 90 cents, the third partner may be 80 cents, and so on and so forth. In a header bidding solution, every demand partner has the same floor and must bid at or above it. It is important to note that a CPM floor applies to the demand partner as a whole, not an individual advertiser.

It is up to the demand partner to find the right advertiser who can fill at or above the respective CPM floor each time it requests a cent. That means that if I'm the publisher, and I have maybe Pandora as we see up here and I have a floor of $1, and media math is the DSP that I am working with, and it's trying to buy the inventory. Media math must find an advertiser that they're working with who is willing to pay a one Dollar more CPM. Now also, when you're on the supply side need to think about optimizations. The goal of a supply side ad ops team is to get the most revenue for their inventory, ie the best combination of fill and eCPM. As we talked about before, there are a lot of different reporting metrics we look at including win rate and fill rate, meaning that for every impression that you send every available impression you have not every single impression will get filled, there may not be demand for every impression.

In the digital space, there is nificantly more supply than there is demand. So you're going to be hard pressed to find 100% fill rate. So let's walk through an example of what that might look like. And if I'm publisher, I may have a million impressions available, and I set a floor of $1. Turns out my fill rate is 10% and My eCPM is $1. We'll keep it at $1 for easy math, that will deliver me a revenue of $100.

So for every 1 million impressions I have at this floor, I'm getting about $100 in revenue. Now, let's say I'm on the ad ops team, and I say, well, that fill rate is pretty low, I want to try to increase it. Well, one way to increase your fill rate is to drop your floor, that means more advertisers will have the opportunity to bid on your inventory. So same situation, again, a million impressions. This time I've dropped my floor all the way to 10 cents. My fill rate has jumped up to 50%, but my eCPM has decreased as well.

So in this situation, even though my fill rate has gone up, I'm only generating $50 per 1 million impressions. So this is not working for me. Let's try another combination. million impressions. Now my floor is 50%. I see my fill rate has gone to 25%.

My eCPM is 50. Since at a 25% fill rate, which means my total revenue 4 million, or 1 million impressions is $125. Obviously, that is the one we want to go with. Now, this process is sometimes automated, but a lot of times it is gone through trial and error to find what is the best combination of floor and fill in eCPM to get you the most amount of money for your impressions on the supply side. on the demand side, the advertisers need to determine how much money each individual user is worth to them, and then set a proper bit. So each advertiser must determine how much a user is worth to them and what kind of data is required to properly identify that user on the exchange.

So let's look at some examples. Nike, Nike may be promoting $125 pair of shoes, they only want Want to promote it to males in California, each customer is worth $90. to them. It's a good value. But because we have a fairly small audience only males in California, and probably something like shoe shoppers are going to need to use third party DMP or an internal DMP. So they're going to need to add some additional data to their bi they're going to have to spend more on it.

So this may be a costly bit. Dunkin Donuts on the other hand, maybe promoting a 599 meal deal with a national reach male female, very little data is needed because their reach is almost the entire country. But each new customer to them is only worth $4 nificantly less than Nike, but they don't need as much data in order to identify those people. So Nike may be willing to bid up to $9 for their CPM because they need a lot of data. They have a limited supply and the value of that customer is fairly large. However, Dunkin Donuts may be only willing to bid $1 50 because they can get their inventory anywhere, nationally, male female, not a whole lot of specifics on what apps they want to run on.

So they may have the same budget, but significantly different cpms. So when we're talking about a bidding environment, this is where we get into a situation of who is going to deliver the highest bid. If you are an app, and your user is using you in California and Nike, decides that that user fits into their target demographic, then Nike is going to weigh outbid everybody else. But if Nike says, No, this person really doesn't fit what we're looking for, but Dunkin Donuts is available, then Dunkin Donuts may be the only one available to win, but they're gonna bid at obviously a much lower rate.