Strip Malls And Santa Claus

During this last quarter of 1999, we will see more online shopping properties than have ever been on the web to date.

You’ve got,,, and CBS Storerunner.

This holiday season will serve to “prove the model” for online real estate committed exclusively to ecommerce.

In a recent ShopNow,, iMall, and Yahoo, to name a few of the biggies. As the industry matures and the rush is on to take advantage of this current upcoming holiday season, other sites, some of which have been around a while, have entered the fray. They are:

All of these sites charge differently. Some require slotting fees for storefront placement, some charge on CPC, some CPM, some cost-per-lead, and some will do a combination of all of the above.

So, how do you go about evaluating all of these opportunities?

At the outset, shopping site opportunities should be evaluated the same way any media placement should be: Can the site satisfy numerical objectives based on past performance results or performance data as provided by the site? As it will be a cost per acquisition metric that will serve as the determinant for success, so, too, should that metric be used for the purposes of predictive modeling.

Quantitative – Numerical Goal Consideration Factors

  • Cost-per-click. $1 or less is an acceptable CPC for any kind of placement.

  • CPM. For shopping placements, assume a 1 percent CTR and a 3 to 5 percent conversion rate. If a proposed CPM against these assumptions meets or beats the cost-per-acquisition objective, go for it.
  • Traffic. A site should be pulling in 500,000 uniques to really be a factor.
  • Amount of inventory. I think it doesn’t make sense to purchase impression levels that are greater than 2/3 the number of unique visitors a site gets on average. So, if the site were getting 500,000 unique visitors, we’d purchase 333,333 impressions. This allows for enough frequency to ensure decent “reach” (I use that term loosely) but not so much that we saturate an audience.

As conventional wisdom for direct response motivated advertising indicates, response happens sometime during the first 3 exposures (I’ve seen an average of 88.5 percent of sales/action take place behind one of the first 3 exposures), I don’t like creating a situation where a site is overbought and we begin wasting impressions that can no longer result in action.

Qualitative – Strategic Placement And Site Architecture

  • Placement. The advertiser’s category needs to be no lower than a second-level category behind a home-page placement. What is meant by that is, if on the front page I have “Entertainment,” the next level down needs to be a category germane to my client’s product or service for the placement to really drive activity. It has been my experience that for each additional level a visitor has to drill down, as much as a 50 percent drop-off in response is experienced.

  • Reciprocal linking. All offered placements must in some way link back to the client site. If it does not, it must link directly to a transaction page that offers the client’s product with the client’s branding. A simple SKU slot is not enough, unless it is treated as a cost-per-click placement.
  • Site recognition. How well known is the shopping property? If it is new, to what extent will it be promoting itself? These are important questions to ask of the newer properties for the obvious reasons: Shopping sites are primarily volumetric plays. Shopping sites do not have as their primary tenor and thrust a relevance to an advertiser’s product or service. Shopping site placements are really more like vying for shelf space for SKUs. If the site isn’t doing the traffic, just like any bricks and mortar, there will not be the volume to make it all worth while.

Shopping sites are a must-buy for a site with an ecommerce component. But they do not always pay out. The guidelines as outlined above should serve as a simple roadmap for quickly determining which sites are, at the very least, worth moving to the next step with in negotiations.

And here endeth the lesson.

Related reading