We know that product recommendations are good for your bottom line — they can comprise up to 31% of e-commerce site revenues. But if you don’t understand why they’re working, you can’t capitalize on their full potential.
Customers are not making the majority of their purchasing decisions just because you recommend an item. They buy recommended items when those recommendations offer value to them, come at the right place in their customer journey, and make their user experience easier.
How and why this happens varies across products and business models. What adds value for a subscription service is different from what adds value to the product page of a clothing site; what adds value to a product page isn’t what adds value at checkout.
Here are five ways to do personalized product recommendations, from why each one works to where the recommendations should be placed on your site.
1. Best Sellers
The “best seller” label says something about the products underneath it — they’re trusted, they’re popular, they’ve got clout with other customers. And the closer you can match your best seller recommendations to a customer’s preferences, the more they’ll trust those recommendations.
But when we say best seller recommendations, it’s important to personalize them. Plugging best selling items without personalization might be a good way to get people into browsing on a home page, but when it comes to nudging customers to buy, you have to be smart.
Take Barnes and Noble’s discount email alert. They aren’t just prompting any old books at 30% off, they’re prompting young adult fiction. For a frequent YA-buyer, these books are right up their alley, and they feel the books are being recommended by Barnes and Noble’s other frequent YA-readers:
If this customer got a 20% off email alert and it was filled with biography, they probably wouldn’t click through to purchase.
The same principle holds for other items as well. If a customer has clicked on a fruit tea page, try recommending best-selling fruit teas. If they purchased hot weather running gear at the beginning of summer, nudge them with an email about the best cold weather running tights.
Targeting your best sellers helps you get the most out of their weight. With just a little information, you can help customers discover more of your best-loved items, bringing them back to your site and getting them thinking about other purchases they can make from you.
Who should use best-seller recommendations: From food to cosmetics to tech, best seller recommendations can have a place, when used thoughtfully.
Where should best-seller recommendations be placed: Because best-sellers’ strength is giving an overview of popular products, best-seller recommendations play better in emails and as site subpages than they do as sections on a product page or checkout, where recommendations should be tailored to items customers have navigated to.
Quizzes are a great way to help personalize recommendations online, and they work best with e-commerce products that are so individual, they are hard to shop for without trying on in-person. Think glasses, which are hard to choose without trying on, or scents, which can vary from nose to nose.
What a quiz allows you to do is put the customer in the driver’s seat. People who are shopping for this type of item online want to find what works, and likely have done research before going in to purchase. They do not want to browse endless options, struggle with search filters, or hunt for product specifications.
A great example of a product quiz comes from skin care and makeup company, Clinique. They have an extensive quiz that asks users about their concerns to provide custom recommendations. Note that they explain why they ask about factors such as sleep, acting as a consultant might in-store:
They also provide an overview at the end of the quiz, to show shoppers how they’ve been charted, and allow shoppers to change their responses as appropriate:
These are factors that raise the degree of confidence a consumer has when purchasing a product. The experienced consumer knows they’re being given a product that fits with their concerns, and new customers are directed to products that work for them.
Quizzes are also a way to get a consumer invested in your product. After going through a multi-question quiz, a customer has sunk time and energy into your product. They have more purchase intent than a casual browser, and expect to be rewarded with custom recommendations.
Which companies should consider quizzes: Companies that sell products or services that are highly individual, such as skin care, makeup, perfume, glasses, running shoes, styling boxes and meal services.
Where quizzes should be placed: Quiz CTAs play well on home pages or on a page all to themselves. It is unlikely that a shopper would start taking a quiz if prompted while browsing search or product pages, as quizzes are a way into product recommendations divorced from other site activity.
3. Custom Subscriptions Recommendations
Custom subscriptions generally start with a quiz or questionnaire that determines the customer’s likes, dislikes, price range, taste, style, measurements, or whatever else is relevant to the subscription. Take shave service Billie’s quick and visual quiz, for example:
Their baseline information gathering is simple preference choices about colors, products and delivery frequency. It’s a short quiz, which makes it easy to get started with the service.
But as a Billie customer –– or any customer of a subscription service –– continues to receive products, you learn more about them. Perhaps they purchase extra products on the site occasionally, or send feedback through an email check-in. Maybe they change frequency or price range on their subscription.
All of these are data points that make for a rich customer profile for subscribers, which means that you can give personalized product recommendations based on their history. If Billie notes that subscribers with once-a-month delivery in the Midwest are disproportionately likely to purchase shaving cream in the winter, they can recommend that product to a subscriber who fits that profile.
If a style subscription box notes exactly what items customers keep and what they send back, they can build a good idea of a customer’s style. As that customer continues to receive their style box, or special offers and product nudges associated with the service, those recommendations can become more and more personal.
Where should recommendations from custom subscriptions be placed: Wherever subscribers can make changes to their subscription is a great place for custom recommendations — think of it as a checkout page that subscribers return to. However you regularly communicate with subscribers is a good place as well, such as email.
4. Visual AI
Giving personalized product recommendations based on a customer’s browsing can and should be better than it is right now. Many product page recommendations and search suggestions rely on general customer information to prompt users to continue shopping — and this doesn’t always reflect user intent, especially for clothing, shoe, and accessory purchases.
For example, a “We think you’ll love…” section for a clothing site with hundreds of items might rely on purchase information that says people who buy flannel shirts also buy jeans. The problem with that recommendation is that you don’t know that any given customer who clicked on a flannel shirt is also shopping for jeans, you only know they are shopping for a flannel.
Rather than cluttering a product page with vaguely related items, using visual AI can give customers more options they are actually looking for. Take Atterley, which shows a customer looking at brown puffer coats an array of similar coats:
This recommendation helps users comparison shop by offering an at-a-glance survey of prices and styles of similar items, which can help them make a purchasing decision without having to leave a product page and go back to search.
The best part about visual AI is that it works equally well to personalize recommendations for every customer, whether or not it’s their first or fiftieth time on your site, whether or not they’re logged into their account or not. It caters perfectly to user intent on every product page, every time.
Who should use visual AI: Clothing and accessories retailers, home furnishings retailers, especially marketplace sites that aggregate thousands of items across vendors.
Where should visual AI recommendations be placed: Product pages and “quick view” options are the best places for visual AI recommendations, as they help users search and comparison shop.
5. Customers Also Bought…
While using aggregate data on product recommendations is not appropriate for some scenarios, it can be a great way to personalize recommendations in others.
For example, think about the last time you purchased a cell phone. If you went into a store, the store clerk probably offered you a number of accessories to complete your purchase: a phone case, a screen protector, an additional pair of headphones, a longer charging cord, etc.
These additions make sense to recommend –– either in-store or online –– because a cell phone is a purchase that usually demands new accessories. Rarely would the case of an old phone fit its replacement, and so people purchasing a new phone will get the new case.
A great example of this in the e-commerce space is grocery delivery service FreshDirect. FreshDirect offers combination purchases when you hover over items, like baby carrots and hummus:
Items that are frequently purchased and repurchased, such as food or cosmetics, give a better picture of how customers are purchasing items in tandem. And, when the cost is low, customers are likely to throw in an item to try out, such as a $4 hummus, especially if they’re trying to meet a minimum delivery.
Contrast this to a clothing scenario: if you are shopping for a flannel and $60 jeans are recommended, the cost of just throwing in the item takes a lot more to justify.
Who should use “Customers also bought”: Clothing and accessories retailers, home furnishings retailers, especially marketplace sites that aggregate thousands of items across vendors.
Where should “Customers also bought” recommendations be placed: Product pages and “quick view”options, in addition to the checkout page. Especially when low-cost items are being recommended, the checkout can be a good way to recommend items customers may have forgotten or might be tempted by.
Recommend with care
Giving personalized recommendations to e-commerce customers is not a one-size-fits-all process. By focusing on the customer intent, the customer journey, and the products that customers are shopping for, you can choose the right method of recommendation for your e-commerce site.