“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 by Andrew Casale, VP Strategy, Casale Media.
In broadcast, there’s a direct correlation between the time an ad airs and that ad’s relative worth. If we’re talking prime time, prices are high, and we tend to see Ford, Coca-Cola, McDonalds and other blue chips battling it out for share. But as the night wears on, viewers instead see their advertiser compliment switch more to local ads running at discounted price points. Higher demand and associated prices are well justified in key day-parts, ratings are high, and there’s a certain cache that comes along with rolling an ad in a coveted slot.
The influence of time in programmatic, at least so far, doesn’t work the same way – not at all. Rather, the outcome is quite unique to the buying channel itself. Many of the most “expensive” time periods are quite contrary to the broadcast parallel. Prices in RTB marketplaces aren’t set so much upfront by sellers (save floors). Instead they are directly set by the volume of liquidity in any given marketplace.
When an impression put up for auction sees a dense volume of bids against it, clear prices go up. When density drops, so do clear prices – makes sense, right? But taking the basic premise of time into consideration, and looking at clear prices based upon time, we start to observe a few interesting dynamics at play. For example, at the start of every hour, bid density tends to immediately spike. That spike in bid density is followed by another commensurate spike in average clear price. The clear price declines immediately following the first minute in an hour, and declines further into the remaining minutes of any given hour.
To illustrate this graphically, we’ve indexed the average clear price that we measured in the first three weeks of January (Jan. 1 to Jan. 21, 2013), by minute. This data is aggregated based on the bid activity we see from the exchanges operated on our platform. For the purposes of more easily illustrating this phenomenon, we plotted each minute of all hours and days with the first minute in the center of the chart. What we see is that on the average, on the 0th minute of every hour, prices spike approximately 7% against the average clear price from the minute prior.
It’s not difficult to rationalize why this might be happening. Most bidding algorithms need to be tuned based on the spend caps they are governed by. Time plays a key roll in getting that balance right and can make or break the difference between delivering evenly, on time, and under delivering against a contract. It would appear that when any given hour flips, a good majority of bidders push a burst of bids through in the first minute to either catch up for under delivery in the hour prior, or “get ahead” so to speak in the new hour, in an effort to help guarantee that they are pacing comfortably against their goals.
It would stand to reason that if we see this kind of behavior each time an hour flips, that we would also see it each time we roll into a new day. We do, and in that case the effect is far more pronounced, marking the first minute of any given day as one of the most expensive programmatic time periods in the day (take that, prime time!)
Beyond the pretty charts, this validates the belief that as this buying channel benefits from continued increases in liquidity driven by additional demand, price averages will rise. We are already clearly seeing this when we “zoom in” to areas of a marketplace that are either highly supply constrained or in high demand. Further, while we remain in a state where supply outstrips demand, this knowledge comes with a (temporary) potential perk to buyers. If you bid less in the 0th minute, you’ll probably save a little bit of money. The inverse perk to sellers is if a buyer is spending more for your impressions in the 0th minute, the perceived value of your media is higher than bid averages suggest.
Follow Casale Media (@casalemedia) and AdExchanger.com (@adexchanger) on Twitter.
