Regular readers may remember my post a few months back commenting on an interesting DataXu article about high price volatility in the RTB-driven spot market for online advertising. They put up another really interesting article Monday.
It seems their research has shown that (what I will call) “overtargeting” is bad for performance.
Regular readers will certainly remember my frequent commentary on this issue in the context of driving offline sales. Many targeting techniques sound hyper-precise (great!), but don’t deliver the goods when impact on key attitudinal indicators like purchase intent, or more importantly impact on offline sales, is measured. Better sales tools than success tools indeed.
Imagine my interest as I read this new research showing excessive targeting doesn’t even work for the DR metrics they studied. If this stuff doesn’t work for branding or DR, what does it work for? Does it serve any other purpose than as the shiny object that helps burgeoning hordes of venture-funded media sales people to convince increasingly overwhelmed media buyers to further fragment their media budgets?
Clients: Don’t take privacy risks or pay premiums for targeting that doesn’t deliver results. Listen with a skeptical ear. If a targeting tactic sounds too good to be true it almost certainly is. The internet is a fantastic marketing channel, but it’s not magic. Anyone that suggests otherwise is either a fool or takes you for one.
A very interesting post on AdExchanger today covering the first installment in what DataXu expects to be a monthly series of reports on market trends.
DataXu disclosed historical volatility of prices across the landscape of biddable online ad inventory they saw through their DSP platform – billions of impressions across multiple exchanges – between 4/10/10 and 5/10/10. The figure? 102%. That’s huge pricing volatility on an absolute basis and (as they point out) much higher volatility than we see in Goldman Sachs share prices, oil prices and even presidential approval ratings.
Broadening the aperture doesn’t change the picture. For example, 102% is much higher volatility than we typically see in historical data for a wide variety of exchange-traded commodities – themselves a notoriously volatile asset class. I think the way DataXu has calculated volatility for online ads may even understate the difference. From the notes on the source post here, it looks like DataXu is calculating this 102% number by measuring the variance of average daily prices within that month period. Volatility in financial markets is usually expressed in annualized figures (like this). I’m not sure exactly where DataXu got the other figures they list, but since annualization is very common for these types of figures I wouldn’t be surprised if those are annual figures. If I am right, then the apples:apples annualized figure for online ads would likely be much higher.
Either way, there’s simply no question that the spot market for online ads is tremendously volatile.
So it has always struck me as odd that the same large manufacturing companies that have active, sophisticated, futures-based hedging programs for raw materials like oats and soybean oil with 30-40% annual price volatility would tolerate volatility many times higher in purchases of online media – an increasingly critical raw material input. As I have written previously in a pair of articles in Ad Age and Ad Exchanger, I think this will change and indeed must change for online media to become a greater share of overall media spend for key categories like CPG.
But in order for that to happen, we need to give them better tools. We need a Futures market.