AI in retail: Increase revenue with product recommendations

Personalized Shopping is nowhere near a new concept in retail, and it has evolved significantly over time, but what once was a competitive advantage has today become a hygiene factor, vital for business success.

Published in

BuiltOn

5 min readSep 26, 2019

According to a study published by cmo.com - 66% of participants said that a lack of personalized content will stop them from making a purchase. This number is alarming, especially with conversion rates for online retail sales already being as low as 4%-5%.

A way to combat loss of customers and automate personalized content is with AI-powered recommendation engines. Once implemented, this technology can understand customers better and deliver targeted product recommendation without big budget increases or time-consuming processes.

In this article, we will explain the specific benefits of AI-powered product recommendation, and briefly touch upon the technology behind it. Finally, we will discuss what is the best way to acquire a recommendation engine.

What is a recommendation engine

Recommendation engines are essentially algorithms. They process data to paint a picture of who your customers are and what they want to buy.

With a clearer picture of what your customers want, a recommendation engine can suggest products more relevant to what they search for. This has a positive impact on the bottom line through up-sales and cross-sales. How exactly do recommendation engines make the product suggestions? There are two ways:

Source: https://towardsdatascience.com/what-are-product-recommendation-engines-and-the-various-versions-of-them-9dcab4ee26d5

Collaborative: This method is about collecting information about past customers, and predicting what future customers similar to them might purchase. For example, both Jon and Judy like cats, artisan coffee and Star Wars. When Judy buys a particular brand of coffee, it is likely that Jon might like it too.

Content-based: This method focuses on product recommendations that are similar to a user’s past preferences. For example, if Jon bought a Star Wars t-shirt before and he’s looking for a new mug, he is more likely to get a Star Wars mug rather than a regular one. It’s also possible to use a hybrid approach, which is used by among others Netflix.

The streaming service makes recommendations by comparing the watching and searching habits of similar users (collaborative filtering) as well as by offering movies that share characteristics with films that a user has rated highly (content-based filtering).

Business benefits of recommendation engines

As touched upon in the previous part, the end goal of implementing recommendation engines is to increase revenue by delivering customer want or value.

What are some of the direct benefits a recommendation engine brings? Here are a few we find important:

Higher customer satisfaction: Consumers are flooded with choices every day. Personalized recommendations cut through “the noise” reducing the time it takes to find an item of interest. Customers also enjoy the discovery element when browsing through items that have been “handpicked” for them.

Increased loyalty: People like to be recommended things that match their preferences. When an e-commerce site can relate to their choices and needs, they are bound to visit that site again.

Increased conversion: A purchase confirmation paired with recommendations of matching products and services, or a “recommended for you” email make customers more likely to buy more.

Generate more cross-sales and up-sales: Recommendation engines also excel at offering suggestions for complimentary products. For example, a matching lens, case or extra batteries when a customer is buying a new camera. In terms of up-sales, features like “What customers are buying now” or “Other customers also view/purchased this” can make a customer aware of higher-end products.

Incorporate new products more efficiently. Researchers at Penn University looked at what impact recommendations engines have on sales of new products and discovered that lack of ratings had little say in sales. If a product was recommended to a customer by the recommendation engine, the probability of sale was high.

Recommendation engines — build or buy?

Once you’ve decided that a recommendation engine is right for your business, the next question is “How do I acquire one?”.

There are two main choices. You can either build one in-house, or partner with a company that offers a recommendation-as-a-service platform.

The final decision rests on:

Your timeline. Functional recommendation engines are difficult to build. If acquiring a recommendation engine is urgent for your business, partnering with a company that has the expertise in building recommendations engines is advisable. If time is not of the essence, consider building one in-house.

Need for scalability. Does your product catalogue change frequently or is expected to grow in the future? In that case you need a recommendation engine that is scalable and is able to handle large sets of data while retaining relevancy. This adds more complexity to the recommendation engine and thereby more time to build one.

Your budget and availability of in-house knowledge. The two go together because building a recommendation engine from scratch can get very expensive. It often requires a team of skilled data scientists and programmers to both build the recommendation engine and maintain it afterwards. With a generous budget, you might afford to hire an in-house team. However, if you are working with a limit budget, you might consider partnering with a Baas company.

The ultimate decision will be determined by a mix of the factors mentioned above. If time and money are an issue to your business, it’s better to partner with a company that has already build a recommendation engine.

Wondering where to start? Schedule a meeting with our team to see how recommendation engines can fit your business.