Here is why. It is long, but once you read it you'll have a much better understanding of how our advertising works.

When you have a traffic surge, advertisers aren't always prepared for it, and so they won't necessarily send through top paying campaigns consistently throughout the surge. 

Advertisers typically commit themselves to buying a particular amount of premium campaigns on any given site on a day. On a day when you're having a traffic surge, they may burn through those faster, and then lesser paying campaigns start to filter through. They'll always be above our floor, but it kind of works like this:

If Kraft has 1000 impressions they need out there today, no matter what, they're willing to pay dearly for it. They'll win our auction for those 1000 impressions, at top dollar. But after those 1000, they might be done for the day, or they might want some more, but they're not as desperate to have them filled, so they're not willing to pay as much. 

These are natural fluctuations to preheader, or auction, bidding. When you're having a traffic surge, you burn through the premium faster, and move on to the "normal". It's all about what the advertiser has planned to spend for the day, and how much inventory you have available.

Programmatic advertising means there's always inventory available. But the CPM becomes very dependent on how desperate the advertisers are to get it in front of your audience. Since we make them bid against each other, they have to prioritize how and where to spend money, and if they've already decided what they're spending on a particular day when you're seeing a surge, they may not pick up the slack the way you'd expect them to.

If your traffic graph has a lot of peaks and valleys - those surges are, in the eyes of advertisers, somewhat unexpected. Unless they're sustained for several days, the advertiser doesn't have a plan that accounts for the higher number of impressions they have available to them.  

That is why you see your RPM go up when the increase is sustained over several days, or you see the RPM go up if the traffic goes back down. They start sending higher paying campaigns to more of your traffic, or you have less traffic that gets a higher share of premium campaigns. Either way, income goes up in relation to pageviews, and therefore, the RPM will go up. 

This is all very natural to programmatic advertising. Correct me if I'm mistaken, but I think we're the only company out there that gives you a daily breakdown of RPM. That's why this is so glaringly obvious, and might seem like something is wrong. 

But the same behavior happens with any network, and any site, and you just aren't being given a view of it. Our own sites show almost identical graphs. When we have a single traffic spike that doesn't sustain itself for several days, our RPM dips in the graph too. It's completely expected behavior. 

In a traditional setup, with a direct sold situation, when the ads were gone, they were gone. They'd fill your ad spots with house ads instead. Or, flip over to the backfill, which basically had the same effect - a lower paying ad in the same spot. 

The nice thing with programmatic ads is that you'll never fill with "house ads" that don't make you any money, or need to make multiple calls out to multiple networks, just to get your ad spot filled. And programmatic advertising with an auction pays much, much more than a waterfall.

If your posts are optimized, and your site is loading quickly, that's really all you can do to make sure that you're giving the advertisers (and your readers) exactly what they want. The rest is down to what purchases the advertisers have planned to make for the day.

Over time, as you grow and have less fluctuation, your RPM will stabilize a bit more because the advertisers will have a better idea of what they should be buying on your site each day. 

NOTE: If your Google Analytics profile isn’t in EST, this can cause a major offset in your RPM as well. Our earnings are in EST so if your GA isn’t, things won’t sync up correctly, and this will be especially amplified during a spike. See this article for instructions on updating your Google Analytics profile to EST.

Did this answer your question?