One Size Does Not Fit All
Walking down the street today, I couldn’t help but notice a sign from the Gap advertising buy one get one free on shirts, tee-shirts, and sweaters. It’s a pretty compelling proposition to have you stop in the store, and the economic implications are pretty interesting as well. The Gap generally has a volume strategy in place with many of their items. Their collared shirts for instance are $25 a piece or $20 each if you buy two or more. They were $20 each today, and then it was also buy one get one free, or as it rang up $20 for two shirts. Imagine though, coming into the store, excited to buy a shirt and not finding your size. That certainly happens in some sales, where the desirable sizes go quickly, but imagine coming into a store like the Gap and every shirt said "one size fits all" on the tab? It can work for some hats and accessories, but it certainly wouldn’t work for a clothing store looking to sell a lot of items. By offering different sizes, The Gap and others make more money, because they are able to offer in effect the right product for the right customer.
The Gap and several others in the clothing business, do something else clever as well. They don’t just offer you different sizes, they offer you different colors of the exact same item. Lacoste and Polo do this especially well. Enter a Lacoste boutique and you will find a variety of polo shirts in a wide variety of colors, all around $70 per item. It costs the company very little to produce different shades of the shirt. The cost is in the set up of the cut, not the replication of it. And, it’s a way to make the most out of having to set up the different sizes of that cut for the original shirt. For the clothing company’s, they succeed by selling as many units as they can at the right price. But, even The Gap, can’t find a product for everyone. That’s why they have Banana Republic for instance. There are styles and prices that would fail if sold at The Gap. And Banana Republic can only do so much, that too is why The Gap also owns Old Navy. If we think about a demand curve with multiple products and prices on it, a company makes the most when it can hit as many points on the curve as possible, and that’s just what The Gap has done with its brands.
In the offline world, some people call this type of segmentation "de-averaging." It’s not a widely used term, especially in our industry, and doing a search online reveals only a thin literature set. But, it’s a term that apparently almost everyone in Capital One knows and a principle that allowed them to significantly grow their direct mail business by essentially making sure they did what a savvy retailer would. For the retailer, it means product segmentation – different sizes, styles, and maximizing the templates for each. For a company that has a relatively limited product set (and one where lots of them don’t necessarily help), e.g., a credit card, the equivalent is being able to offer an optimized experience for each consumer. Doing so, though is no easy feat, and for them it meant leveraging massive amounts of data and creating models so that they didn’t send the same piece of mail to every user. Equally important, it meant being able to vary the product pitch and details of the product as well – they focused not just on the creative but the entire funnel by user segment. Hence the term "de-averaging" since they are moving away from just a single process / flow. Were they just to send out a single flier that they then tested and tweaked for maximum performance, that would be an improvement, but it would still be playing to an average.
A handful of online companies are starting to think about this de-averaged approach towards their marketing, and it should have significant applicability for the performance marketing space. Ours is a world all about customer acquisition, but if we think about how most people market today, they only play to the average. If we think about the fake blogs for instance, there is a lot that those running them do that play to the average but also just a little that starts to hit on the notion of de -averaging. Most run the same ads and the same blog for all of their placements. They might make different banners to see which performs the best, and they might test slightly different versions of the landing page. One thing they have done, though, is try to personalize the page by reading the IP address and having the fake blog author from that city, e.g,, "I’m a mother of three in <city name>." All of a sudden, we have a different experience for different users. What they really should do, is take that approach and try to create different banner, landing page combinations by placement.
The savviest take things a step further, and it is why we are on the cusp of a breakthrough in online marketing. Instead of just creating different banners and landing pages by placement (size,geo , etc.), which is using data you know, you can now start to tie in data about the user that you didn’t already know. A whole knew breed of companies exist to help augment user data on a cookie level to append it with learnings about that user, from purchase intent to types of sites they have visited. In addition, and this is especially true of the free trial advertisers, they collect actual data about the users, so they can then map the successful conversions back to the additional cookie level data they might now also collect. The marriage of data means they can know how not just who to target, but what experience they might want to present to that user. One who has had certain behaviors might get a different story or mix of products than one who has another. The possibilities are almost endless, which is why it isn’t an easy or manual process to solve, but it is one that can lead to incredible gains in efficiency for the advertiser and the end user.
