FREE DM Review Site Registration!
Sign-up today and access DM Review on the Web!

Your FREE registration entitles you to:
FREE email newsletters

FREE access to all DM Review content

FREE access to web seminars, resource portals, our white paper library and more!

   

The Long Tail of E-Commerce and Beyond

  • Jack Jia
  • DM Review Special Report, September 2007

In the last a few years, the Internet has turned commerce on its head. Suddenly, once-sound retail practices, like the mass merchandizing of a handful of popular products (think Cabbage Patch Kids, Tickle Me Elmos, Nintendos or Air Jordans), no longer guarantee a big financial payoff in today's retail environment. The marketplace has become so segmented by niche consumers that bestsellers only produce a fraction of the revenue they did previously, while niche or specialty goods can translate into big business for retailers who learn to tap that end of the market, commonly referred to as "the long tail."

In catering to the long tail consumer, e-tailers have several advantages over traditional retailers, such as infinite shelf space and the ability to more easily change or add to inventory. This allows e-tailers to offer an unlimited selection of high-margin items at a fraction of the cost required of a brick-and-mortar retailer. In fact, it's now possible for online retailers to profit from the elusive long tail by applying social science to their e-commerce technology for the very first time. Moreover, businesses can leverage these same social science principles to improve any electronic interaction they have with a known customer or prospect, especially in the areas of e-marketing and e-support.

This article will examine the different domains of the long tail on the business Web and demonstrate how it's possible to let your anonymous site visitors make highly targeted content and product recommendations that significantly increase revenues.

Harnessing the Wisdom of Invisible Crowds

Matching millions of online shoppers with tens of thousands of products is easier said than done. To do this effectively, e-tailers must think beyond the old merchandising rules, which are too costly to develop, maintain and are not resilient to the changing preferences of shoppers. Scientists have discovered that a random group of informed visitors can predict far better than any individual merchandiser what products people want. This is known as the "wisdom of crowds."

The latest in online merchandizing is now based on what's called crowd sourcing, or using the behavior of the invisible crowd of online shoppers to make product recommendations to one another on behalf of the retailer. Unlike specific product reviews, online retailers use the past context of shoppers who have been to the site to make product suggestions to the visitors that come after them.

In this model, retailers are leveraging the implicit, emergent behaviors of visitors who are anonymous and unknown to each other. By understanding these visitors and their intent, vendors can identify the thousands of microsegments of shoppers who come to their Web site and ultimately match each one with the best products available.

By observing which products truly give value to customers, e-tailers gain an in-depth understanding of community preferences and the thousands of microsegments that emerge around products and categories without the risk of being misinformed by survey bias or misleading click-based data gathering systems. With this approach, the silent majority of site visitors are represented instead of being ignored. In addition, merchants can better understand how these communities self-organize into like-minded peer groups and better serve these micro market segments with more unique products, making the long tail even longer!

For example, an online store discovers that many of its shoppers purchasing kitchen appliances are also looking at flat-panel TVs. The retailer can conclude that these appliance shoppers would be likely to consider a small flat-panel TV when remodeling their kitchens. Online merchants are better equipped to tap into the long tail on their own Web sites because they can uncover the unique, previously hidden desires of their customer base.

Importance of Contextual Recommendations

Tapping into the long tail isn't just about adding more items to the menu. Research shows that shoppers feel overwhelmed and are less likely to make a purchase when confronted with too many decisions. The key is to target a smaller set of products to the right people at the right time.

One school of thought believes the problem can be solved with another round of personalization, profiling and behavioral targeting. The concept is to target products based on individual browsing history together with demographic information.

The latest social science research, however, has proven that this thinking is flawed. As it turns out, individuals have thousands of profiles and past interests. A person can be a father, son, brother, golf lover, traveler, wine drinker, engineer and HR benefit seeker all at the same time. When taken out of context, our past behaviors poorly predict our future. On Amazon.com, cross-product recommendations have seen plenty of misfires outside of book suggestions that worked beautifully. The reason is simple: a book recommendation is within the context of the book. But to always recommend diapers to an individual who's made a past purchase of a baby gift will most likely miss the mark.

Human psychology has revealed something even more profound that we're often not willing to admit - humans are like pack animals and our needs tend to not vary too widely. Given a context, 95 percent of shoppers purchase the same types of products repeatedly. Context is a synonym for the micro long tail segments discussed earlier. By detecting like-minded peers, we effectively discover an unlimited number of buyer segments. Based on the common needs of like-minded peers, e-tailers can recommend products more precisely. And the purchase rate goes up dramatically as a result.

Well-documented research has shown that contextual targeting (a shopper's current context regardless of historical interests) gets 62 percent of the recommendations right while historical behavioral targeting gets it right only 18 percent of the time.

Because context is time-sensitive, recommendations must also be timely, dynamic and taken from real-time feedback. For example, when Valentines Day approaches, the crowd begins surfacing gifts that are common to shoppers for that holiday. On February 15, however, no one is buying chocolate hearts any longer. If the store is still recommending sweets over springtime patio furniture, it will lose business. The recommendation system must be able to detect real-time changes in season, consumer tastes and market trends to avoid falling out of sync with customers.

Soft Landing with Google Context

Recall that your visitors are telling you exactly how to grab their attention on your site. Consider tapping into their experiences to deliver dynamic landing pages for each visitor coming from Google or other traffic referral sites. Rather than creating custom landing pages for every possible natural or paid keyword - an impossible feat, to say the least - merchants armed with an intimate knowledge of their site's invisible crowd can use their collective wisdom to create dynamic recommendations on every page known to convert visitors who have come in through the same query. By doing so, merchants make their sites infinitely more "sticky" and increase sales.

For example, a customer may use the search terms "Fix a broken pipe" on Google and land on the Home Depot site. Her intent is not to look for broken pipes, of course. Showing her glue products together with PVC connectors and a do-it-yourself book will go a long way to serve her needs. By better understanding user context, merchants are able to connect customers to the specific products that like-minded peers found useful.

Beyond E-Commerce

Now, let's broaden our thinking a bit beyond just e-commerce. The connectedness of today's e-life has actually spawned several other tails, in addition to e-commerce. Using a near-identical prescription as above, businesses have the potential to implement a mass-personalization model for any electronic interaction they have with a known customer or anonymous prospect.

Here are a few examples of how the long tail theory and wisdom of invisible crowds can be applied within a number of other business scenarios.

Socially Driven On-Site Search

Your site visitors will punch in an infinite number of keywords into your on-site search engine (a very long tail indeed). Instead of relying on your legacy keyword-matching search engine to find the right content, tap into your wisdom database to display the content links, products and even rich-media assets that have proven to convert previous site visitors when asking the same question.

The wisdom of invisible crowds is also highly relevant for businesses with large e-support practices. Traditional support applications require explicit voting on whether an article satisfies the user's need. This technique results in severe survey bias because it only considers the input of a few users, who typically do not represent the majority, and leaves you with little confidence on how to improve the customer experience. However, your invisible crowd implicitly understands each and every customer interaction, guiding successive visitors to the articles, content and downloads they are looking for as if it were second nature.

Individualized Email

Why send a generic email to your customers or semi-anonymous visitors? By storing a tiny fraction of visitors' context on your site, your email management system can now consult your "wisdom" services to get a highly specific set of products or content links to include in an email that is tailored to that user. Don't settle for a sub-two percent response rate to email.

RSS Feeds

With the wisdom of like-minded peers, you can create your own long-tail of individualized RSS feeds that are highly relevant to subscribers and will keep them reading. Most people read about one in 10 feeds that they subscribe to because they are all generic. If these feeds could intelligently identify the content that subscribers have actually enjoyed, and syndicate that, they would be far more likely to read, enjoy and take action on it.

Mobile

Real estate on your Web site is limited, so you have to use an uncomfortable amount of discretion to display the most targeted set of content possible. In a pull model, where the visitor actively retrieves information from you, ask your invisible crowd in real-time which content is most likely to help the user achieve their goals in the least number of screens. In a push model, handset owners can be pushed content alerts as mobile users "like them" find value in newly published content. This avoids the onerous tasks of defining and maintaining static subscription channels and delivers a long-tail of content recommendations that are highly targeted.

Implicit Folksonomies versus Explicit Taxonomies

Folksonomies, content organization through social tagging, have long eclipsed rigid taxonomies as the de facto mechanism to make content findable in a world of infinite consumers and infinite content. A venerable example of the flexibility of folksonomies is Flickr, and more recently Digg. Fundamentally, the long tail of descriptors that visitors will use to describe your content and products will make them more discoverable and consumable by visitors who are more likely to speak the language of other visitors than your marketing or online merchandising team.

In order to understand the true intent and buyer preferences of your customers, you need to harness the implicit wisdom of your invisible crowd. With this wisdom you can tap into your own long tail - e-commerce, e-marketing, e-media or e-support -and let your customers boost conversion rates for you.


Jack Jia is a founder and CEO of Baynote, Inc. Earlier in his career, Jia served as a senior vice president and chief technology officer for Interwoven Inc. and was responsible for the long-term strategic technology direction of the company. Jia is a frequent speaker at major industry and financial conferences. Jia may be reached at jack@baynote.com.

For more information on related topics, visit the following channels:



Industry Vendors