Technology Management & Leadership

I was reading The Long Tail by Chris Anderson when it dawned on me; the mass customization industry is the incarnation of the long tail principle. Absolute diversity, zero inventory cost, and the difficult issue for customers of finding what they need. The way those products are handled at the moment is similar to a standard catalog of shelf products. We categorize them, define sets of options for individual SKUs, use some cross-selling and up-selling techniques and have consumers go through a lengthy process of deciding exactly what they want. The problem is, they know about some features of their desired products, but not all of them. It’s easier for someone to say: “I want to buy a big, nice, leather book for my wedding party photos” than “Hey, let’s buy a blue faux-leather hard-cover book with a keyhole and stitched binding, silver halide printed high-end paper pages and a feather effect on my pictures!”

We now put them through a series of decisions on issues that may not matter to them, by having them decide up front the exact product they want. Let me reference the research made by Sheena S. Iyiengar about the paradox of choice. In short, the study had a booth where consumers could taste and buy jams. When only six different types of jams were offered, 30% of them bought some. When 24 types were offered however, only 3% bought any. I tend to agree with Chris Anderson, and scope the issue as being the effort associated with making a choice, and the availability of tools to help in the process of making that choice. If we do work in a long tail environment, then the knowledge is there on how to make it work. But first, why do I think we should consider ourself in a long-tail business?

I think there are two evidences. First, the fact that customers want to create a unique product reflects a part of the thinking that allowed the long tail to exist. Second, the current offering is widening as thousand of products and exponentially more unique templates are being created, trying to grab pieces of the tail. More precisely:

1. We have obvious niches, and users move between them. The market of “American wedding photobooks” is a niche as is the “Children birthday cards”. They all have their hits and their tail in their available themes and formats, but they all sell to a different degree.
2. It costs almost nothing to reach those niches, at least for the content variations. The issue we have in our industry is how each distributor and manufacturer controls the formats and the content. The former is difficult to extend, but the latter is simply a matter of standardizing the format and opening the valves. I mean, each customer is actually creating its own content on your behalf for free, and buying it!
3. We strive on correctly categorizing the expanding amount of content and products so users can find them. That’s where better filters and statistics will come handy, but we’ll come back to this later.
4. There is pretty much zero inventory cost for a new “product” for an existing product type and format with content variations.

So, now what? Does this mean we should become Amazon, eBay or Rhapsody in how we present our products? I think we do, actually. We already have aggregators, content producers, fulfillers and distributors, but we don’t have a standardized distribution format, and democratized content production yet. This is worth its own blog post however; at the moment, I would like to focus on what you can do right now within the existing boundaries of your company.

The first concept I suggest might seem counter-intuitive, but I think it makes sense. We currently track product attributes such as size, texture, theme, etc. to nicely structure everything when we display the products to the consumer, to extract useful statistics and to simplify product management. Even though there is nothing wrong in doing this, it is somewhat similar to the jams example. We show a huge amount of products and themes to the user, and we ask them to make a single, enlightened choice. What if we considered, like Netflix or Rhapsody, that every single permutation is a unique and different product? We know something about each product, since we generated them. We can tag them, and we can search them. We have statistics to know which is popular for whom. We don’t need to care that much about what is the single most popular product; we can show the single most popular product for the current user profile, let them decide what criteria are important to them and always make sure something of potential value is being offered to them.

Just throwing a huge amount of products in a never ending list will definitely loose the customer. What we need is:

• Identify who the current user is. What do we know about them? Male, female? Regular user? Lots of photos? Age? Where do they live? Especially, what have they ordered before?
• Build up information on the user. What did they order? What did they like? How do they perceive their “niche”? This can be done by allowing them to “like” themes so they can find them later, “tag” projects so they can find them easily, and especially track which ones they paid for. If there is additional value to the user, some of them will opt-in. For those who don’t, we need to make sure it’s ridiculously simple to simply let the software help you make a choice.
• Show a few items that have the maximum potential value to them. If we are in November, a 38 years old woman from New York logs in for the first time, the product to logically display would be (I’m wildly guessing) a red leather book with a keyhole and a Christmas template. And if we’re wrong? The users will make a choice, and this choice will help the next users make theirs.
• Let the user very easily specify what they are searching for, using faceted search for example. If what is important to them is to pay approximately 40$ for the largest possible book for a Christmas event, let them specify that. And show them the best, specific products matching their choice.
• Acquire feedback from users. Once the order is placed, ask some optional questions that will help you dynamically change the products grouping based on what users want. Even show some of that feedback, either as reviews or simply as an aggregated recommendation: “This product was selected by 75% of our customers searching for wedding books!”

Ergonomics are key to this thinking, since we don’t want user being shown with a huge form with sliders and drop-down menus everywhere. Optimally, we would have an advanced search to please advanced users, but the entry point should make the choice easier, not harder. The platform has to very early drill down the niche the user searches for. It may be an event-based decision, a memories storage decision or simply a way to share images. Once that segregation is done, not only the themes can be sorted and filtered, but also every attributes of the product. We might find out that if an amateur photographer wants to create a book to share his or her work with friends and potential customers, the books to display on the front page will not only be a plain black theme, but also a high-end paper, a printable cover and a reasonably sized format to allow easily bringing copies with them.

This theory needs testing, and I think this is where Google’s approach with live experiments fits very well. The platform should allow for organizing content differently very easily for a subset of customers. See how they react. Are sales better or worse? Are users spending more time creating their books? Do you have better user loyalty? The first step is therefore to determine the scope of the category. All customizable products, photobooks, print products? Then, implement and iterate in an empirical fashion. Then, determine what is required to make the it scale through the industry. Standardized content format? Content aggregators?

Did you have an experience that supports or contradicts my argument? I would love to hear your feedback.