Overcoming Disintermediation

Woe, the beleaguered middleman.

Not long ago he was the dominant member of the supply chain. Then, like a meteor crashing through the earth's atmosphere, along came the Internet, dramatically changing the climate, threatening to drive him into extinction.

In less apocalyptic terms, it's called disintermediation and it describes the elimination of the middleman--stockbrokers, travel agents, traditional retailers--in the channel of distribution as producers abandon distributors, wholesalers and retailers because technological innovations have allowed them to sell directly to end-users, enabling them to reduce the number of intermediaries in the selling process.

But the belief that disintermediation will eliminate all middlemen is flawed. The foundation of disintermediation is the ability of the consumer to go directly to the producer of a product or service. The consumer is then limited to the products or services of that single producer. Ultimately, this drawback will prevent the extinction of the intermediary. The true benefit of the intermediaries of the future will be their ability to create value by customizing a product or service solution. Intermediaries will base this customization on intimate knowledge of consumers' wants and needs. They will satisfy those needs with products and services from multiple producers.

Reintermediation Through Customer Insight

What we really need are more middlemen, albeit new kinds of intermediaries, that add customer value. One way to effect this "reintermediation" will be for middlemen to add significant consumer value through in-depth consumer insight. This will give intermediaries the ability to personalize the consumer experience and recommend products and services from many producers in real-time.

As intermediaries adjust to changing business models, new types of companies are emerging. For example, "one-stop shop" middlemen provide dramatic business efficiencies with sophisticated recommendation, personalization and targeting capabilities. Such new intermediaries are changing business relationships, frequently shifting a company's focus toward the customer and away from the product or service.

Who are some of these successful companies? Much has been written about Amazon and Yahoo! They are succeeding because they act as middlemen who add significant customer value. What is Yahoo! doing if not mediating between the consumer who wants something and the myriad of places where that thing might exist? Technology has been the driving enabler of this capability. As Michael Roberts states in the Educom Review, "In the longer run, perhaps the greatest contribution of the Internet and its related technologies to society will be its ability to bring organizations and institutions closer to the individual." The ability to add such value exists at every customer touchpoint--from the Internet to the brick-and-mortar point-of-sale to the call center--provided the right technology is used.

The Recommendation Engine

Using consumer insight to personalize customer experience and, based on that knowledge, make targeted product and service recommendations, is at the foundation of reintermediation. Today, hundreds of Web sites, as well as an expanding number of call centers and points-of-sale, use personalization. Companies greet the user by name, custom advertisements target the consumer specifically, custom content solidifies the consumer's affinity to the company, and custom recommendations encourage the consumer to make a purchase. The average Internet-surfer is likely to encounter personalization at retail sites, portals and news providers. In the future, they will enjoy its benefits at most call centers and retail outlets.

How does recommendation technology work? The recommendation and personalization market has split into two different product types:
• rules-based engines and,
• collaborative filtering engines.

While these technologies differ in the way they determine optimal personalized content, they have many similar features. Both solutions seek to understand a consumer's affinities and customize the customer experience accordingly. Both detect user preferences through manual means (for example, user-stated preferences in a survey) and induction (for example, user preferences determined through analysis of their actions and purchases). Finally, both must access information from other databases in order to actualize recommendation and personalization processes.

Rules-based personalization engines use business logic embedded in conditional statements to create content display. Under rules-based personalization, when a consumer's known preferences fulfill certain criteria, the customer is presented with appropriate corresponding content. The system typically uses an interface to input if/then criteria, specifying each condition and the content to recommend in response. The primary benefit of this approach lies in its ability to link the marketing strategy to the customer interaction directly.

...And How It Works

Imagine a fictional music store that implements a rules-based recommendation engine. A consumer visits the store or calls to place an order for a copy of The Three Tenors CD. This interaction provides preference information to the rules-based engine, which has been programmed to categorize each user according to his or her selections and to make sales recommendations of similar items. The engine queries the product database for cross- and up-sell items. It then recommends a collector's edition Luciano Pavarotti CD and interview (up-sell), a portable CD player (cross-sell), and a special 10-disk collection of Giuseppe Verdi operas (up-sell). The customer service representative, store clerk, or Web site then makes the recommendation and, hopefully, closes an additional sale.

Collaborative filtering engines use the recorded preferences of all users in order to find content that is likely to appeal to an individual user. The engine:

• groups consumers by their shared preferences, and through the use of this preferences database,

• predicts optimum recommendations for an individual customer,

• locates people with preferences similar to the consumer's and,

• recommends items that those people liked.

Benefits of Collaborative Filtering

• Responds constantly to changing consumer preferences and ever-increasing information without requiring the creation of new personalization rules.
• Allows companies to target recommendations to smaller groups that might otherwise be feasible with rules-based engines.

Imagine that our fictional music store decides to use a collaborative filtering engine to match customers with content. When our customer arrives online, he's asked to name his three favorite artists. He lists Jimi Hendrix, Janis Joplin and Jefferson Airplane. The collaborative filtering engine then

• searches for people in its database who have similar preferences. Finding a list of several thousand it,

• queries that list to determine their favorite music, and

• makes recommendations.

The next screen the customer sees has a tie-dyed background, a banner advertisement for the new Volkswagen Beetle and recommended merchandise: a Carlos Santana CD and a Woodstock video. As preferences change, the collaborative filtering engine automatically accounts for them and updates its recommendations. Should these aficionados of 60s music begin purchasing Kenny G CDs, the engine will recognize the trend long before most people do.

Such examples highlight the importance of these technological enablers, a point not lost on industry analysts. Some estimate that the emerging online personalization industry will grow to a $5.3 billion market by the end of 2003. But even with the obvious value that reintermediation delivers, is it enough to secure a future for the stockbrokers, travel agents, florists, real estate agents, automobile dealers--even retailers that are at risk? Some may disappear, but many will transform, either by offering more value in the distribution channel, or by evolving into totally new kinds of intermediaries.

CRM Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues