Jul 62010

The Cost of Cannibalism

My monthly column at Search Engine Land, in case you missed it.

Revenue share deals are appealing for their simplicity as well as the sense they give that they protect the advertiser's bottom line. As long as the commission rates make sense financially, it seems the advertiser can't lose money.

However, the fact that some sales are cannibalized from other channels can change the economics significantly if the degree of cannibalism is material.

The reason for this is simple, while the pay-out percentages and discounts may make sense on the incremental sales driven by the channel, the fact that they're also applied to cannibalized sales the advertiser would have gotten anyway make the true costs of the program higher and somewhat harder to see.

Let's take a look at how this works:

Say an advertiser has 50 points of margin on an average order through this particular channel. Suppose that on top of that standard percentage, perhaps a 10% off coupon has been applied on the average order. If we add another 9% variable cost for pick/pack/ship, and some type of revenue sharing commission, the P & L for this program starts to take shape.

Let's say last-touch credit tracking assigns $100K in sales to this program.

On the surface, the numbers look pretty good.

Last Touch Perspective:

However, when we recognize that some of these sales likely would have happened without that last marketing channel, and that those sales came at lower margins and higher costs because of that last touch, the numbers change.

Incremental Sales Perspective:

Here we "tax" the marketing program for the incremental cost of sales (COGs, variable costs, commissions and discounts), and for the costs associated with giving discounts and paying commissions on sales that weren't incremental.

Note that as the incremental percentage drops those other costs pretty quickly swamp the program.


The implications of this for traditional rev share relationships that involve a fair amount of sales cannibalization (affiliates and email) are pretty obvious, but for those who pay their paid search agency on commission the implications are pretty serious as well, particularly for advertisers that do a great deal of offline marketing and hence generate tremendous sales volume on their brand.

Let's take a look at how that model might be structured:

The revenue sharing agreement above is structured such that the agency covers the media costs as part of the revenue share. The incremental percentage torpedoes this program because the revenue share pays the agency for sales on the advertiser's trademark. Obviously, most of those sales would happen through the advertiser's organic brand ads if no sponsored link was present.

Indeed, we've seen too many cases of agencies generating 70 or 80% of their fees off the first 5 minutes of work on the account in setting up the advertiser's brand campaign. It doesn't create much incentive to do the hard work necessary to build a competitive search program, and that absence of incentive to work on the most valuable piece of the program is usually evident when we look under the hood.

Cannibalism can make any revenue sharing arrangement far more costly than it seems. This makes it imperative that parties agree on how sales will be attributed so that dashboards and scorecards remain in sync and that the rev-share percentages makes financial sense all around. That attribution management system must be more sophisticated than simply last-touch gets all the credit because too many orders that end with a search on "Acme" or "Acme Coupon" would have happened without that final touch.

There are other problems associated with revenue sharing, but as more and more marketing programs offer "cost per action" pricing, the problem of cross-cannibalization will get worse.

What seems like a sure-fire pricing model guaranteed to protect the advertiser's interest turns out to be far less attractive under the hood.

If you're interested in playing with the numbers in a spreadsheet, here is a toy model. Enjoy!


5 Responses to "The Cost of Cannibalism"
Dan Merton says:
Hi George, Great article, I ponder this subject often, particularly when it comes to attribution on things like retargeting campaigns where agencies like to take full credit for view-thru traffic. The larger question, for me anyway, is how do you begin to quantify the cannibalization as a percentage? We all know there is some with every program. I know it's higher with retargeting than with PPC, but how much higher? What about affilates, who has you mentioned often get credit at the last min, because because a card in hand shopper, with something in his cart typed in "store coupon code"?
Thanks, Dan. You're absolutely right, not only is it a problem, it's a growing problem with more and more CPA deals being offered as a safe way to test a marketing channel with "no risk". Answering the question "how much of my [XXXX] program is incremental" is challenging, and we think the right "solution" involves a sophisticated approach to attribution management. We've waded into this field precisely because we think there is a need for a good, reasonably priced system that helps advertisers gauge exactly this sort of thing. The answers may not be perfectly accurate, nor without controversy, but having some thoughtful approach to parsing credit beyond last touch is a big step in the right direction.
Billy Wolt says:
view thru's....Sorry, but I lol'ed when i read that. Google tries to tell us how strong our remarketing is performing. So all this time we've had a $20/cpa, and now with the magic remarketing, we are performing at $1.24....amazing :) I like the sound of managing atributions, but I can't see any third party or affiliate being happy with a % of credit being applied (also affecting their bottom line). Maybe I am just old fashioned, but the last click method isn't so bad.
Hi Billy, There is no question that display impressions do create lift in site traffic. The question is how much incremental lift and what did we pay for it. Giving impressions 100% of the credit for sales is ludicrous when the true incremental lift is likely closer to 10%, but they do deserve some credit. Last touch is better than having no system for attribution and giving 100% credit to each program that touched an order, no question about it. Within each channel, last touch makes a great deal of sense as well. However, across channels you'll find last touch gives far too much credit to affiliates, email and remarketing efforts, than more sophisticated approaches would give. George


Check out what others are saying...
[...] This post was mentioned on Twitter by Alan Rimm-Kaufman and others. Alan Rimm-Kaufman said: rkgblog: The Cost of Cannibalism: My monthly column at Search Engine Land, in case you missed it. Revenue share de... http://bit.ly/d58Rnm [...]

Leave A Comment