The challenge of online retailers, now and always, is to sell products that the online shoppers can’t touch or feel, and can’t play around with to make sure that they would like those products. Online retailers have tried to solve this problem in many ways: from product image galleries and product specs to give you more factual information about the product, to customers rankings and reviews for more subjective opinions on the product from fellow shoppers, online retailers try to put potential customers at ease about buying a product that they will pay for now but may only see in a few days.

This challenge is combined with the obvious challenge of trying to sell products to their potential customers. All retailers work very hard to make sure that the products are appealing, that the price points are just right for the market segment that they’re targeting, that the product is easily locatable (in a physical store or on an online website) and a myriad of other factors that come into play when trying to sell a product.

Selling is expensive, which is why when trying to sell products to individuals, retailers usually segment those individuals into market segments and try to use the same selling techniques on all the members of that segment in the belief that the right kind of appeal to the right segment would result in sales. Traditionally segments had been created using demographic information: segments had been defined as people from a certain geographic area that share certain properties such as age group, profession, gender, etc. Marketers would assume that people who share demographic characteristics would be likely to be interested in products for the same reasons, and thus the same marketing and selling techniques would be used on them as a group, and delivered through segment-specific channels like magazines, websites and TV shows that this segments tunes in to.

A newer way of defining segments is based on behavior, also known as psychographic segmentation. Where as demographic segmentation assumes that people behave in certain ways because of certain statistics (location, gender, age, income, etc.) and groups them based on it, psychographic segmentation groups people based on their past behavior. For example, people who buy at Whole Foods would likely also be interested in organic food, tea, environmentally-friendly products and electric cars. Psychographic segmentation works on the assumption that people who engage in the same activities and buy the same products would likely have something in common and therefore can be approached in the same way.

Along with demographic and psychographic information, the web 2.0 era and its plethora of social networks, blogs, forums and other online profiles allows up to gain further insight into people’s buying behavior. It turns out that a very good indicator to what people buy is what their friends are buying. This is a stronger indicator than mere demographics – if all my friends have an iPhone, then I may consider buying an iPhone despite not necessarily being in the market segment that Apple is targeting.

With the advent of online social media aggregators, online retailers now have access to their customers’ social graph. A sophisticated online retailer can use this social graph to determine, for each customer, who their friends are and what they bought from the same retailer or other retailers, and provide individually-tailored suggestions to each customer. This is basically friend-based merchandising, and is a very powerful tool: if all your friends are reading the same book, buying the same phone/car, eating the same food, you, as a customer are far more likely to at least try that product. Merchandising based on customers’ social graphs could be a very interesting trend that we may see in use in the upcoming years.

However, social-graph-based selling and merchandising has its limitations:

• Privacy concerns: Facebook tried doing this with their Beacon initiative, and suffered a backlash from their users due to privacy worries. Retailers who will attempt this must be very cautious in the manner in which they present this to their customers.
• Scope limitation: There is currently no automated way to share purchase history between retailers, so while very large retailers might still find the technology useful since at least some of their customers’ friends have also bought from the same retailer, small retailers would probably find that their customers’ friends have probably not visited them yet and thus friend-based merchandising would provide no insights. It is possible that retailers would collaborate to create a huge data repository of what all of their customers bought, but it is likely that this move will cause a big backlash from the public and quite possibly the government.
• Audience limitation: This technology relies on people using such technologies as social networks and blogs, and having their friends also use them and be online shoppers. People who use such technologies are usually younger and web-savvy, and contain very few aging and elderly people. Thus, social-graph merchandising (“friend-based merchandising”) would work mostly for retailers targeting younger audiences. As the US population is growing older, retailers will have to be wise about utilizing this technology to avoid data bias, and to perhaps encourage older shoppers to create their social graph in a 65+-friendly environment.