Rethinking Retargeting

Just a quick post to make sure folks saw Richard Frankel’s article today on AdExchanger.

A couple solid, related tidbits in there.

The first point is about attribution.  Richard cautions that retargeting often “steals” attribution from other tactics unless careful steps are taken to prevent it from happening.  DR tactics stealing attribution from upper-funnel tactics is an important topic on which we have written before.  As we mention in that article, it’s also the subject of an entire body of work by Microsoft’s Atlas Institute.

The second point is about the importance of finding new prospects and customers, not just retargeting old ones.  This difference between “demand creation” and “demand fulfillment” (as’s Jim Spanfeller has somewhat famously put it) is something that needs to be understood and carefully considered when developing a comprehensive marketing and media strategy.

As online media marches past a 30% share of total media consumption, new technologies are eroding offline media like TV and print.  Both demand fulfillment and demand creation budgets alike must follow consumers online.

Misconceptions about Yield Management and Channel Conflict

Another interesting article from’s Jim Spanfeller yesterday.  I wholeheartedly agree his point about the online ad industry focusing too much on demand fulfillment and too little on demand creation, as evidenced by my previous posts here and here.  That’s exactly why we built from the ground up — to help advertisers with demand creation.  I also agree with his point about ad networks that offer some types of user-based targeting  representing a potential “data drain” and a legitimate privacy concern for publishers.  This is an important issue and just coming to the fore for the publishing community overall.

That said, I disagree with two major points Jim makes in this article.

First, he seems to be perpetuating industry confusion on the definition of “remnant”.  In the context of online ad inventory, “remnant” is commonly considered to be the opposite of “premium”, which is often used interchangeably with “high-quality”.  Thus if “premium” = “high-quality”, then “remnant” = “low-quality”.  Unfortunately this is often untrue.  When used correctly, “remnant” actually means “available to the spot market after forward commitments have been fulfilled”.   So the opposite of “remnant” is not “premium”.  The opposite of “remnant” is “reserved in advance”.  There are really two distinct axes at work here:  one describes quality of the inventory, while the other essentially describes the terms or process under which the inventory was purchased.  There is some correlation between the two axes, which I believe is at the heart of the persistent confusion; it’s a fact that remnant inventory is often of lower average quality than inventory that is reserved in advance.  However, due to traffic volatility, forecast errors, suboptimal pricing, supply/demand imbalances, etc., there is often significant volume of high-quality or “premium” inventory available in the “remnant” market. The airline standby example he cites is actually a good illustration of the correct definition of remnant, not (as I think he suggests) the incorrect one that has done so much mischief.  The standby seat has exactly the same physical characteristics (“quality”) as the seat sold in advance, but the difference in timing and deal structure results in a difference in value to both the airline and the passenger, which manifests in a difference in price.

This brings me to my second point: Jim’s position in this article on airline yield management practices shows some pretty fundamental misunderstandings.  Airlines’ lack of profitability has a lot more to do with unions, over-capacity and sub-optimal product offerings than it does customers risking their vacation plans or business objectives to save money on a last-minute ticket.  So I would echo Jason Kelly’s well-informed comments on the thread and add that to suggest yield management practices are somehow to blame for the poor financial performance of airlines is like suggesting that ERP systems and supply chain optimization practices are responsible for the poor financial performance of the American auto industry.  It’s simply not true.

The bottom line is that ad networks and publishers can work together for mutual benefit over the long haul, but to do so requires careful management of channel conflict, an issue we take very seriously.  This discussion is a valuable and important one, but I think we need to be more careful and rigorous in our thinking – the more so, the better off we’ll be as an industry.

%d bloggers like this: