The chemistry of great CX: Great recommendations

The bar for customer success has never been higher – when it comes to online shopping, customers expect excellence and fast service.

Coveo’s Ecommerce Relevance Report 2022 found that 93% of consumers expect online shopping experiences to be equal to, if not better, than in-store. However, 94% find their online shopping experiences not always relevant to their buying habits and preferences.

Providing an always-relevant shopping experience to meet customer satisfaction is a tall order, but not impossible. The key lies with an effective ecommerce recommendation strategy. 

According to McKinsey, recommendations are responsible for 35% of Amazon sales, and 75% of Netflix selections. Much like how knowledgeable sales associates are sought after by shoppers at brick-and-mortar stores, e-commerce stores need a digital equivalent – a smart recommender – to win purchasing customers. But most importantly, recommendations must be 1-1, personas no longer cut it. This is incredibly difficult to do when 80%+ of shoppers shop anonymously.

There is a formula for creating a smart recommender - like chemistry, there are elements that make up a successful ecommerce recommendation strategy. When you combine these elements with AI technology you have the recipe to successfully deliver relevant personalized customer experiences that also benefit the goals of the retailer such as profitability.

The Fundamental Elements of Recommendation

Just like chemistry, which organizes elements by categories in a periodic table, you can have a “periodic table” of essential recommendation elements. In this table, each element can be grouped into five “periodic groups” of recommendations: Context-based, business-based, profile-based, product-based and content-based. Let’s break them down:

Context-Based Recommendations

Context-based recommendations suggest products based on product attributes, shopper’s context, observed trends, and other people’s behavior. 

Using information such as geography, behavioral data, historic data, and more, you can recommend trending products, best-selling products, new products, top-rated products, and style recommendations. 

The benefit of this type of recommendation is that it works with anonymous shoppers. 

Business-Based Recommendations

The goal of business-based recommendations is to suggest products based on business priorities. In short, not only should your recommendations be oriented towards generating profit, satisfying supplier relationships, and flushing out expiring inventory, but they should also align with your brand image.  

When you hear “business-based,” think of suggestions like on-sale, curated recommendations, and sponsored recommendations. They are built using data related to pricing, transactions, inventory, return rate, spending, and others.

Profile-Based Recommendations

Buy again, recently viewed, and affinity-based suggestions all belong to profile-based recommendations. 

Incorporating suggestions that offer products based on a shopper’s profile and preferences is another way to solve the cold shopper problem – meeting the expectations of anonymous shoppers while respecting their desire for privacy. 

This suggestion works by looking into demographics and personal data, explicit data provided through surveys and registration forms, as well as implicit preferences inferred from clickstream and other clues. 

Product-Based Recommendations

Online customers also like to see products related to what they’re looking at, which is where product-based recommendations come in. Giving options such as buy again or recently viewed is another helpful tip in getting shoppers to purchase more.

Such suggestions are not possible without having data like product metadata (price, images, categories, etc.) or behavior-based data (e.g. items most commonly viewed together) to craft similar products, gradient products, and complementary product recommendations.

Content-Based Recommendations

Suggest non-product content to inspire buyers, facilitate buyer enablement, and support customers with content-based recommendations. 

This type of suggestion includes inspirational-enabled recommendations, inspirational content recommendations, and self-service content recommendations. 

The Next Step: Combining Elements for Success

The real science is combining all of the elements to develop an effective ecommerce recommendation strategy. Of course, to do so, you will need a recommendation engine that can unify all the different types of data, and an intelligent AI layer. With machine learning you can create a “just-for-you” shopping experience - at scale - that wows your shoppers. Merchandisers are given the tools to allow them to hypothesize, experiment, and to test new strategies. With this in mind, any business has the opportunity to meet and exceed customer expectations while simultaneously driving revenue.