Editor’s note: T.J. Gentle is the CEO and president of Smart Furniture, Chattanooga, Tenn.
Personalized experiences engage customers. That principle has been well-known to merchants for hundreds of years – maybe even longer – ever since they discovered that learning the names of customers and greeting them personally resulted in repeat business. Learning customer preferences and showing them merchandise that appeals specifically to them results in greater loyalty.
That’s a fairly simple concept, but the idea of 
personalization is still emerging in the world of e-commerce. Personalization almost always results in increased engagement. However, when interaction is happening between a person and a computer, the challenge faced by retailers is much larger than hiring a smart and friendly sales staff.
The rewards of besting that challenge are great. Web shopping experiences enhanced through prescriptive personalization have been shown to increase conversion rates twenty times, increase revenue per visit by a factor of 22.9 times and increase average time spent on site by 700 percent. And that’s in the current environment. Over time, retailers who fail to personalize their sites will start to see their numbers spiral into decline.
Though personalization is in its infancy, there are already several methods being explored. Choosing the right method is important to optimizing its impact.
Know your audience
Netflix was an early adopter of online personalization. The digital video provider hastened the decline of physical video rental stores by providing quick and easy search for mail-order DVDs and on-demand digital streaming content. But it could not duplicate the experience of browsing shelves at a local shop or getting recommendations from employees. Netflix addressed this shortcoming by providing suggestions based on what customers watched and how they rated the content.
When a Netflix customer rates a movie, an algorithm using a “wisdom of crowds” methodology looks for other customers that rated the movie the same way and then makes recommendations based upon what those customers rated highly. Even with something as subjective as media tastes, this crowd wisdom logic works well for Netflix.
Amazon uses a similar premise for the “you may also like” feature, which provides recommendations based on what customers who follow similar click paths ultimately purchase. This form of personalization works for these retailers because it adequately simulates the experience customers would receive in a physical setting. Simple, qualified recommendations are perfectly suitable for movies and consumer durables.
When to get prescriptive
Things get more complicated with products that customers spend a great deal of time researching and expect knowledgeable retailers to help them navigate. Few people start the search for a product like furniture with a clear idea of what brand, style and color they want. They visit stores and look at what appeals to them, talking to sales staff about their preferences, budget and expectations. Successful stores hire salespeople who listen carefully to customers and lead them to products likely to appeal to them and fit their needs.
Since online shopping journeys are entirely self-led, this puts e-commerce at a disadvantage. A personalized shopping experience cannot be effectively duplicated with a “wisdom of the crowds” approach because:
- stylistic taste is only one of several components that guides a purchase and
- the complexity of products offered requires expert knowledge to create meaningful recommendations.
Achieving personalization for furniture product categories relies on robust customer and product knowledge equal to what a seasoned salesperson is capable of delivering. The best way to extract customer preferences and match them with product attributes using an algorithm is through prescriptive personalization – combining expert product knowledge with information about what individual customers prefer.
Profiling customers
In a brick and mortar retail environment, sales staff can learn about customers by engaging them in conversation and asking questions, as well as looking for body language and other hints to help guide the experience. Web sites have not traditionally engaged visitors in this way, but it is possible to ask customers questions and store their answers in a database tied to their accounts. This is the first step in prescriptive personalization.
By offering an incentive, retailers can draw customers into taking online quizzes that classify them into a specific customer segment. On SmartFurniture.com, customers can take a short “this or that” quiz that simply asks them to make a selection from three images. After answering just ten questions – which most customers can easily do in less than two minutes – an individual profile is created that indicates stylistic preference, budget, size constraints and lifestyle.
Classifying products
Once customers are properly segmented, the personalization algorithm can begin matching them to products most likely to fit their needs and wants. To do so, however, requires exhaustive quantification of each individual product’s attributes. Building this data set requires a team of experts to break down each product and assign numeric ratings to features such as style, build quality, materials, size, etc. Accuracy of matches between customers and products increases with the level of granularity associated with product attribute quantification.
Because they are based on specifically targeted questions and expert analysis, the intricacies of the customer and product profile databases that result from prescriptive personalization eclipse those that are possible using a wisdom of crowds approach. This creates the specific, meaningful product matches that are appropriate for larger-risk purchases and those for which a curated shopping experience is expected.
The evolution of personalization is advancing to machine learning, with algorithms that are capable of analyzing the relationship between shopping journey click paths and conversions. Paired with available data from inbound referring links and IP information, it is likely that fully curated prescriptive personalization experiences will become possible without the need for customer quizzes in the near future. That level of precision will further increase shoppers’ experiential expectations.
Online personalization is becoming more scientific and easier to institute, with technologies that are more reliably duplicated than brick and mortar personalization. The time for static Web sites has passed and the path forward is through highly personalized Web-based shopping.