The Many Shades of Website Personalization

Imagine if you could have a real-time greeter on your website, ready to ask visitors about their interests and directing them to the right set of products. No doubt your conversions and average-order value would go up.

Obviously that’s not possible, so e-commerce sites must rely on technology to gauge visitor interest, suggest products, and even make personal recommendations based on a consumer’s demonstrated preference.

Personalization can be as simple as A/B testing, or as sophisticated as conversational marketing fueled by artificial intelligence. The more sophisticated the higher the costs, but those costs are likely to be offset by an increase in conversions and higher average order values (AOV).

The right solution will depend on your budget, size of product category, schedule and skillset. You can always upgrade to a more advanced form of personalization as your website grows.

Segmentation vs. Personalization

Vendors in the field of website personalization are particular about the distinction between segmentation and personalization.

Segmentation is a rules-based approach of discovering groups of customers with a common, yet broad set of characteristics, such as geo-location, IP address, time of visit, mobile operating system, and so on, and groups them into segments. Next, marketers analyze the shopping behaviors of each segment (e.g. people in New England purchase winter boots in December). These insights are used to create a custom homepage featuring winter boots for website visits with an IP address that indicates they live New England.

This is distinct from real-time personalization, which delivers content based on the unique characteristics and behaviors of each website visitor (if the customer has a history of purchasing pink clothes, the website will present pink snow boot options). The more the website owner knows about the individual shopper, the more personalized the content.

Types of Website Personalization

A/B Testing

A/B testing is making decisions on what the new baseline should be based on personalization tests. Sometimes tests are based on random selections, but more likely they’re based on other factors that are driven by customer segments. For instance, if you identify your high value customers and push specific only to them, that’s personalization.

A/B testing can deliver stronger conversions and higher average order value (AOV), especially when used with audience segmentation. You can A/B test your homepage and product category pages to see which delivers the most clicks, conversions and higher AOV among your customer segments.

A/B testing is relatively easy to implement, especially if you use a solution like Optimizely, which automates the entire process. These solutions automatically divide the traffic, and stop a test once you’ve reached a statistically significant result. They also make it very easy to compare results side-by-side.

Product Recommendation Engines

Broadly speaking, there are two approaches to product recommendation: collaborative filtering and content-based filtering.

Collaborative filtering recommends products based on what similar items consumers looked at or purchased previously. It assumes that consumers who’ve liked or purchased similar products in the past will like similar products in the future, and that those purchase patterns will be relevant in the future. It’s algorithms collect and analyze behavior and predict the products most likely to be of interest to a consumer based on that consumer’s similarity to other users in his or her cohort. These predictions are specific to the consumer, but are based on information gathered from many consumers. A benefit of collaborative filtering is that it can be applied broadly and can recommend a wide range of products accurately.

Content-based filtering is built on product descriptions as well as user profiles.  Algorithms recommend items to individual consumers that are similar to those they purchased previously. The product information can come from the product description (e.g. women’s waterproof winter boots). User profiles are developed based on items they’ve searched, viewed or purchased. Because they’re based on measured interests of each consumer, content-based filtering systems are highly accurate.

In reality, most engines offer offer a hybrid of both approaches.

1:1 Website Personalization

Website personalization customizes every aspect of a consumer’s website experience based on the unique visitor, from the homepage, product category pages, and even real-time promotion messages delivered to the user while onsite. Recommendations aren’t static, as real-time behavior is used to inform the website experience. For instance, a consumer may have a long history of purchasing red shoes, but if she is currently viewing black dresses, the personalization platform will recommend black shoes from a preferred brand, rather than red ones.

Artificial Intelligence: Conversational Marketing

Conversational marketing is a futuristic approach to personalization. It’s a solution based on artificial intelligence that can simulate human-like conversations. Chatbots are leveraged to automate conversations between a website and a visitor, and every conversation is recorded and analyzed. AI can also detect patterns that aren’t easy for human beings to detect, and then use those patterns to personalize the website to each visitor.

Interesting Vendors

There are a variety of vendors with platforms that offer a range of personalization, from A/B testing, to product recommendation and conversion-rate optimization. The vendor you choose will depend on your budget, of course. We recommend selecting a vendor that offers a wide array of options, so that it may evolve with as your revenues grow.

  • Adobe Marketing Cloud: Most mature product and highly suited to enterprise-class sites. Its personalization suite is the de facto standard. The product leverages Omniture Sitecatalyst product, the original high-end analytics suite.
  • Monetate: Monetate Intelligent Personalization Engine has a strong focus on 1:1 personalization, across all consumer touchpoints.
  • Optimizely: Optimizely allows you to customize your entire website experience. It started out as a platform for A/B testing, and has expended from there.
  • Evergage: A real-time personalization platform, using machine learning.
  • HiConversion: Started out in the multivariate space, and do a lot of work with demandware. This company is interesting because its products integrate with Magento.

Have questions? Liked what you read? Let us know!

Written by: Phillip Jackson, Ecommerce Evangelist