Arbitrage of Ad Impressions
Arbitrage is one of the most fascinating concepts in digital advertising over the past several years and still exists today. The term represents the practice of the simultaneous purchase and sale of an asset to profit from the exploitation of differences in price between both identical and different markets.
There is CPA arbitrage, in which one purchases a lead for $5 and sells it immediately for $8 to a buyer who is in need of it.
There is CPC arbitrage, in which one purchases a click from a traffic provider to a certain page and sells that visit to the owner of the page for a higher price. Similar to a broker.
Then there is CPM arbitrage, which involves buying an impression early in its life-span and selling it either in the same exchange or different exchange for a potential higher price. This is possible only because of the speed of auctions happening in fractions of a second. CPM arbitrage occurs in display as well as video, desktop as well as mobile; it is just that video auctions take at least twice as long.
Refer to Dhar Chart Fig. 7.
If you take a 150 millisecond timer, you can hold 1–6 auctions before the final ad-creative is served or the impression disappears. Thats the strange thing about arbitrage, even if the impression times out before there is an actual creative served, the publisher, exchange-seller, Network-D, Network-C, and Network-B all made money. It is Network-A that suffered the loss because it purchased the impression without having enough time to sell to the Exchange-Buyer.
The scale of arbitrage is not mentioned or discussed in industry articles and trade publications because the majority of people are not aware of the extent to which it has reached. Until recently any exchange buyer could purchase 100’s of millions of impressions a day at around $0.01 CPM and resell these in other exchanges for an average of $0.05 CPM. This profit of $0.04 CPM seems small initially but if you take 100M impressions at $0.01, the cost is $1000. The revenue is $5000. The gross profit is $4000 and once the ad-serving fees and rev-shares are applied, the net profit comes to around $3000 for little to no work which remains essentially on auto-pilot until the dynamics of the market demand change.
Although this may seem malicious or even bizarre to some, the only reason these opportunities existed was because there were end-buyers placing bids of $0.02-$0.25 CPM in the exchanges. The campaigns attached to these low bids were meant solely to fill budgets and thus had very minimal targeting requirements outside of the “users” being in the US.
Arbitrage is the dark side of daisy-chaining and waterfalls because it is done by parties who do not own or even officially represent the ad-space in question. Most arbitrage transactions are not connected to the final buyer or original seller, but are between various arbitrage players that simply re-sell without adding value. The negative effects of arbitrage must NOT be confused with the “tech-tax” described in my previous post, although many of the similar platforms make incremental revenue along the way.
Although this is a fascinating concept and innovating method of revenue generation for the individual party, it corrupts the value of the impression and creates a larger than necessary disparity between buyer cost and seller revenue.