Skip to content
Brittany Dec 17, 2021 14 min read

Recommendations best practices part one: Building a 360° strategy

This blog is part of a series about product recommendations best practices. Check out the other instalments here:

Here's a simple truth: Recommendations work. But here's an equally simple truth: They don't always work. 

Many studies have shown the power of adding personalized, typically AI-driven recommendations to ecommerce sites, with stats like:

  • 75% of customers are most likely to buy from a retailer that recognizes them by name and recommends products based on previous purchases
  • Personalized product recommendations account for just 7% of visits but 26% of revenue 
  • In 2020 businesses lost $756 billion because of poor ecommerce personalization
  • Shoppers that click a product recommendation are nearly 2x as likely to come back to a site 
  • 60% of consumers say they will likely become repeat buyers after a personalized shopping experience
  • 49% of consumers said they've purchased a product they did not initially intend to buy after receiving a personalized recommendation

Still need convincing? We've rounded up an entire collection of stats that detail how customers are asking for personalization and what real-world impacts marketers are seeing to their bottom line and retention rates by implementing it.

Once you understand the value of personalized recommendations, the next step is to actually implement them. It's easy enough to find a recommendations engine on major platform markets like the Shopify App Store or BigCommerce marketplace.

But installing a recommendations tool without a strategy is kind of like giving someone flour, yeast, sugar, and oil and asking them to whip up a loaf of French bread - without a recipe. In both cases, without guidance, training, or strategic thinking, the task may be doable, but the results probably won't be great. 

If you're going to invest in personalization on your store (a key trend that's not going anywhere), it's critical to build out a plan first. Luckily, we've got you covered with our guide to creating a 360° recommendations strategy. 

The basics: Not all recommendations are created equal

One of the most common missteps when it comes to implementing personalized recommendations is not understanding the role each type of recommendation plays in the buyer journey. All recommendation types can be effective, but only if they're deployed in the right way. The goal of any recommendation or personalization should always be to serve up the right product at the right time to enhance, not distract, from the buying experience.

Ideally, your recommendations strategy should serve up content across three levels:

  • Personal level
  • Product level
  • Store level

We'll dive into what types of recommendations suit each of these levels, but the most important thing to recognize is hanging your strategy on just one of these categories won't work. A 360° recommendations strategy involves tapping into all three of these areas.  

Keep reading to learn about:

  • What recommendation types exist in each category
  • When and where to deploy various recommendation types
  • Best practices and strategies for different recommendation box types 

Personal level recommendations

When most people think of personalization, they assume recommendations are generated on a 1:1 basis, just for them. And while most recommendation box types adapt to customer buying and browsing behaviors, the reality is truly personal-level recommendations are primarily driven across just a few types.

What distinguishes a personal-level recommendation in this definition, is they are 100% driven by each shopper's individual behavior, whereas product and store level-recommendations take into account more macro inputs. 

Why you need personal-level recommendations

Personal-level recommendations are designed to make it easy for shoppers to keep track of where they're at in their shopping journey. They may not be the most compelling on their own because they're directed by the shopper themselves, but they are essential to provide structure to anyone's shopping experience.

Recent Views

If there is one recommendation type every site shouldn't be without, it's Recent Views. Think of these boxes as bookmarks or placeholders, as shoppers either browse other products or leaves a site entirely. Recent views allow customers to pick up where they left off.

There is no AI involved in recent views either. It's simply best practices to have them available so a shopper doesn't have to hunt for something that caught their eye, speeding up the time to hitting checkout. 

Best practice: Place recent views at the bottom of virtually every page of your site. Experiment with home page placement; higher up may lead to more conversions as you shorten the time to value for returning shoppers.

 

Recent Purchases

These boxes function almost identically to Recent Views, in that they're 100% customer-generated and 0% AI-generated, but with a focus on items a customer has recently bought. 

Not every site is suited for showcasing recently purchased items, however. Recent Purchases are mostly only applicable for brands that have a replenishable element to them, such as vitamins, beauty products, or food. Customers appreciate not having to hunt down the exact size or formula or flavor of an item they've previously bought, making it easy for them to add to cart - again! - in a single click.

Still debating? If your store doesn't sell replenishable products, but you think a customer wants easy access what they've bought in the past, to say, recommend to a friend or post a review on, chances are Recent Views can cover this need. 

Best practice: Add Recent Purchases to the customer account center - where they're most likely looking at past orders to begin with - and in re-engagement emails to keep customers' favorites top of mind. 

 

You May Like 

There is one type of personal level-recommendation that involves AI, and that is You May Like recommendations. As the title suggests, these recommendations serve up recommendations based on a customer's actions and behavior. Specifically, You May Like presents related items based on the shopper's recently viewed products, in real time.  

Best practice: If a customer has yet to click around on a site, the You May Like box will default to Popular Items instead, ensuring the highest potential for clicks and conversions. 

Product-level recommendations

Things start to get a little more interesting once we start looking at product-based recommendations. Drilling down to a product detail page (PDP) is a clear signal a customer is interested in something in particular. Recommendations and personalization provide an opportunity to either provide the next step in a customer's journey, or increase their order size. In some cases, it may actually do both of these things. 

Product-level recommendations, like personal-level recommendations, can involve a mix of both AI-driven and manual approaches. In both cases the primary goal is to drive shopper interest and ideally, purchases. 

Why you need product-level recommendations

Imagine walking into a store with a single rack of clothes. After browsing that one rack, you'd probably feel like you're done and walk out the door, right? This is why most stores implement a grid or herringbone layout, giving shoppers smaller sections to explore so they're not overwhelmed, or conversely, underwhelmed. 

Product-level recommendations are a virtual form of merchandising, giving customers a logical 'next click' much in the way a brick and mortar store's layout has an organic flow. If anything, however, product recommendations are even more effective than traditional retail. While a physical space might group together 'party wear' or 'office wear', virtual merchandising can get even more specific by honing in on the exact details of a particular product to suggest similar or complementary items. 

This not only improves the customer experience, it keeps them engaged and on a site for longer - and ideally, with more items added to their cart. 

Frequently Bought Together

When most people think about recommendations, they often visualize Frequently Bought Together picks. This style of recommendation was famously popularized by Amazon, which credits 35% of their revenue to their recommendations algorithm. 

These boxes are powered by AI that takes into account store-wide data about, quite simply, which products customers tend to buy at the same time. While Frequently Bought Together boxes are most often found on PDPs, they can be placed anywhere by setting a 'reference product' that reflects an item a customer has in their cart, or even one they've bought.

Best practice: Frequently Bought Together recommendations can be presented as a traditional carousel, but they can also be shown as a bundle which customers can add to their cart in a single click. This method works particularly well for brands with smaller inventories or items that are truly meant to go together, like film for a camera or a brush for a beauty palette. 

Related Items

The difference between Frequently Bought Together and Related Items comes down to a matter of strategy and the type of inventory a store carries. Frequently Bought Together boxes show items that are often bought together as a set, while Related Items highlight similar items or alternatives. Related Items are useful in case there's something that's not quite hitting the mark for a customer; instead of clicking away to a new site, they can instead look at something else in the same vein. 

It's possible to make a case for having both types of recommendations. For example, if you were browsing a menswear store and looking at a dress shirt PDP, a customer may be interested in both Frequently Bought Together items (a suit jacket, tie, or trousers), as well as Related Items (other dress shirts in the same fabric, by the same designer, or the same color). 

Best practice: Consider setting a reference product to present related items in other places on your site beyond a PDP, such as the cart, or even checkout. 

 

Upsells and Cross-sells

First, a quick overview of the difference between upsells and cross-sells, as most people tend to use 'upsell' as a catch-all term for both tactics. Upsells refer to making more money off of a single product or purchase, by encouraging customers to buy a bigger size, a more premium product, subscribing, adding on features, or adding on services, such as a warranty or installation. A cross-sell, in contrast, is about generating more revenue by encouraging customers to buy multiple products at the same time, like a matching set, accessories, or refills. 

In both cases, the common goal is to boost cart sizes. And in both cases, the upsell or cross-sell is specifically linked to the product a customer is showing direct interest in. AI can be used to automatically curate both upsells and cross-sells, although in many cases, it may make more sense to manually curate them. For example, a fashion retailer may want to encourage customers to 'Complete the Look' and present cross-sells as a bundle. Or a water filter brand may present an upsell that entices a shopper to buy two filters to get 50% off the second one. 

No matter what, remember the golden rule of upsells and cross-sells should be that they're creating value for the customer; not feeling like a 'cash grab'.

Best practice: Don't be afraid to spotlight upsells and cross-sells in a prominent way. Use bold styling or embed upsells and cross-sells in pop-ups to give customers the opportunity to fully consider your offer.

 

Store-level recommendations

The final category of recommendation types is the least personal, but no less important to your overall strategy. Store-level recommendations often take into account macro store trends, and are useful for giving customers a direction to start or expand their shopping journey.

Unlike other categories of recommendations, store-level recommendations are designed to push a shopper's journey beyond what they may specifically show interest in. The objective is the same - keep customers clicking and shopping - but the focus is much less personal to help expand their horizons. 

Why you need store-level recommendations

Think of your store-level recommendations as the brick and mortar equivalent of an attractive window display, or a 'new arrivals' section. They're useful to catch the eye and give customers a quick glimpse of what a store is all about. This is particularly helpful for people who are net new to a site, giving them an easy first click. Alternatively, customers who have landed on a site to browse, as opposed to hunt down a specific product, might appreciate having a birds' eye view of whatever it is the store wants to promote. 

Most Popular

When it comes to store-level recommendations, Most Popular is well, one of the most popular choices. This recommendation box aggregates all store data over an extended period of time to spotlight what products are perennial bestsellers. 

Most Popular boxes are generally best employed by brands that have a relatively stable stock of products - for example, a protein powder brand or a tea shop. For brands that are constantly introducing new SKUs or cycling through seasonal trends (as in fashion), tracking most popular items can be a little trickier. 

Best practice: One of the ideal spots for Most Popular items is the home page, to help guide new shoppers on what your all-time beloved products are. But Most Popular recommendations can also be placed on collection pages, to highlight what items are top-sellers in different categories; particularly useful for businesses with large SKU counts. 

 

Trending

While Most Popular boxes showcase all-time bestsellers and favorites, Trending boxes focus more on what's happening right now. Depending on a site's traffic, Trending boxes will showcase the products that are 'going viral' over the past 24 to 72 hours, based on views, clicks, and purchases. 

Trending boxes work well for brands that refresh their stock pretty often, such as beauty, fashion, books, or electronics. Give customers a quick glimpse of what's new, hot, and worth their attention by adding Trending boxes to the home page or collection pages. 

Best practice: Put a twist on 'New Arrivals' page by presenting Trending recommendations as a grid. The hottest products will continually update on the page, driving the highest probability of conversions as sales beget sales. 

Featured Collection

Sometimes it just makes sense to show customers specific products. Enter Featured Collections. Use any collection as the basis for spotlighting a showcase of specific items. If you're using a platform like Shopify, it's possible to create unlimited collections using smart rules that keep the collection up to date.

Consider different angles like price point, color, material, interests, weather, gender and more when curating collections, then spotlight them on various pages to keep customers from aimlessly clicking through the nav.

Best practice: A featured collection doesn't have to be called a featured collection. Label it anything - like 'Our Favorite Gift Ideas Under $50', 'Back to School Essentials', or 'Get Set for Summer' - to give customers some context while they browse. 

 

New Arrivals

The final type of recommendation is basically a subset of a Featured Collection. New Arrivals highlight what products have freshly landed on a store, giving new and returning customers an easy first click to drop onto PDPs. 

Best practice: While it's tempting to have customers click into a full collection page of the newest products, having a New Arrivals carousel on the home page can be an easy way to signal to customers you actually have new stock, and give them a more digestible glimpse of what exactly has dropped. 

Why all three types of recommendations are required

Let's break this down. Each of these recommendations types serves a specific purpose for shoppers, and in many cases, helps to move them along their shopping journey.

New customers or those who haven't visited recently will benefit from store-level recommendations. These recommendations entice shoppers to see what's new, trending, or beloved, giving them confidence to click into a PDP to learn more.

Once they're on a PDP, product-level recommendations are a useful tool to create more value for the customer, by suggesting other products they might like (while boosting cart sizes). This helps the customer either feel more self-assured in the product that first caught their eye, or excited about rounding out their order with additional products that complement the object of their affection. 

Finally, you've got personal-level recommendations. These recommendations provide structure to the shopping experience, particularly for returning shoppers that have yet to convert. They make it easy for shoppers to backtrack or pick up where they left off, while giving their experience a personal touch. 

This is why it's so critical to not stop your strategy at one or even two categories of recommendations. A comprehensive 360° recommendations strategy considers all shoppers and where they're at in their buying journey, to help move customers along in an organic, positive way. 

Looking for help in building your recommendations strategy? Contact LimeSpot to book a demo, or try us for free.  

REQUEST DEMO

 

Sources

  • 75% of customers are most likely to buy from a retailer that recognizes them by name and recommends products based on previous purchases (Kinsta)
  • Personalized product recommendations account for just 7% of visits but 26% of revenue (Salesforce)
  • In 2020 businesses lost $756 billion because of poor ecommerce personalization (Accenture)
  • Shoppers that click a product recommendation are nearly 2x as likely to come back to a site (Salesforce)
  • 60% of consumers say they will likely become repeat buyers after a personalized shopping experience (Twilio)
  • 49% of consumers said they've purchased a product they did not initially intend to buy after receiving a personalized recommendation (Segment)

 

avatar

Brittany

Brittany is LimeSpot's Enterprise Marketing Manager who loves all things ecommerce, optimization, and writing - making her the perfect guru to walk you through even better ways to use LimeSpot's personalization suite.