Meanwhile, discover strategies to boost conversion rates, transaction size, and customer engagement in our Idea Guide.
Product recommendations: It's LimeSpot's specialty yes, but it's also becoming increasingly commonplace in the ecommerce industry to offer them.
Yet for as familiar as product recommendations are, there's still plenty of questions about just how they work. Is it sorcery? Well, not exactly. While there's certainly some behind-the-scenes technology that goes into serving up the right product to the right customer at the right time, as we'll show you, the entire process behind product recommendations is actually pretty simple.
In this blog post, we're breaking down some of the top questions we get about product recommendations to help demystify the magic. Do you want to grow sales on your Shopify store? We'll help you become better at understanding and optimizing the presence of product recommendations on your site, answering the question once and for all, what is a product recommendation engine, exactly?
Product recommendations engines consume data from activities that happen on your ecommerce website - typically related to browsing or purchasing behavior - and analyze that data to find commonalities and relationships to serve up personalized recommendations.
A data-driven product recommendation engine should analyze and present product recommendations across three planes. First, by understanding the relationship between related products - for example, this coffeemaker is often bought with this type of coffee pods.
Secondly, good recommendation algorithms take into account product data through what's known as natural language processing. This type of analysis understands the characteristics of a product to better understand a similar product that is like it, or alternatively, a good match for it. Natural language processing might consider what product category an item belongs to, or what the description is for a specific product to infer details about who or what that product is actually suited for.
Finally, the smart adjustments behind your product recommendation engine should also be able to infer a shopper persona. It's important to recognize that a shopper persona doesn't actually have to reflect a customer's own demographic profile. For example, if a customer is browsing pink sparkly velcro-strap shoes on a shoe website, the machine learning might infer a persona of a young girl - even if it's actually her mom doing the browsing. As a result, that shopper should be categorized into the 'young girl' persona and served up personalized product recommendations that fit the persona; perhaps other pairs of children's slip-on shoes or rainbow tie dye rain boots. This type of personalized recommendation serves up products a customer might like based on the behavior of similar visitors.
There are many different types of personalized product recommendations, including:
All of these product recommendation types can be organized under these three categories:
Personalized product recommendations are based on the specific actions an individual has made, whether viewing or buying a product. They might include things like Recent Views or Recent Purchases, but can also extend to You May Like recommendations.
Product level recommendations reflect relationships between products, as understood by your machine learning. Your algorithm will recommend products based on items that are Frequently Bought Together, Related Items, and Upsells and Cross-Sells. These types of product recommendations are most often driven by intelligent collaborative filtering.
Finally, store-level recommendations are driven by the overall sales data of your store. These can include the Most Popular best seller items of all time or Trending items from the last 24-72 hours. Or they can cover specific Featured Collections or New Arrivals.
As a best practice, it's important to offer all three categories of product recommendations to give shoppers a comprehensive personalized shopping experience with every recommendation type they could ever want.
Product recommendations are based on the sales data and inferred persona of your shoppers, along with the actual product data such as description, title, and category. In addition, product recommendations may also take into consideration the browsing data of a visitor.
Your recommendation algorithms will consume all shopper data to understand the relationship between products and personas. From there, it will work in real time to serve up the right product through dynamic recommendations to the right customer, for the highest possible conversion potential.
Product recommendations are a proven way to drive conversions and boost average order value. By serving up a more relevant product recommendation experience, customers are motivated to continue their shopping journey without needing to search for an item or manually navigate through a site's collection pages.
Instead, they are presented with logical 'next steps', or even first steps, based on the inferred persona of the visitor.
Conversion rates go up with personalized product recommendations because customers are better able to find what they're looking for.
Average order value goes up with personalized recommendations because shoppers are shown more things that they're more likely to convert on, at different points in their customer journey.
Finally, retailers that deploy product recommendations generally see an uptick in customer satisfaction and retention for their online store, largely because the visitor had a smooth, delightful shopping experience that they'll happily return to.
Product recommendations can be used just about anywhere on your ecommerce site. The most common placements for product recommendations including the home page, collection pages, product detail pages (PDPs - AKA a product page), and the cart page. However, product recommendations can also be placed in more unexpected locations, such as at checkout, on the post-purchase thank you page, on a personalized search page, in the customer account center, on content pages, and even in pop-ups or the site navigation.
Savvy marketers have also learned that serving up a recommended product via email is a great way to drive up open and click rates on all types of emails.
Different types of product recommendations work well for different areas of your website. For example, Most Popular or Trending are great choices for the home page, because they can help new visitors understand what is a popular product on your site without needing to dig too deep. Product pages and the cart page are great spots to place cross-sells and upsells based on whatever products a shopper has shown interest in as well. Understanding the best practices for recommendation box placement is a key part of launching product recommendations on your site.
Not all product recommendation engines are created equally. Many recommendation algorithms only consider basic product sales data, without understanding the actual relationship between products.
Top tier product recommendation solutions should offer:
Ideally, you also want a product recommendations engine that guarantees results. Interested in taking LimeSpot for a spin? You can try our product for free for 21 days - our customers typically experience 2-5x conversion improvement.
One of the easiest ways to improve your product suggestions is by investing in an data-driven product recommendations solution. The algorithm behind these types of solutions take into account way more than just product data to form a holistic view of every customer journey through collaborative filtering: Who are they, and what are they most likely to buy?
It's also a good idea to AB test your product recommendation widget placement. Different merchandising strategies in terms of where the product recommendations are placed both site-wide and on a single page can go a long way to driving more conversions and bigger basket sizes.
It's important to have an overall personalization strategy in place when adding product recommendations to the mix.
The easiest way to measure the success of your product recommendations comes down to core ecommerce KPIs:
In addition, you may want to look at the difference in metrics between shoppers who interact with product recommendations and those who do not. Typically speaking, we see an AOV increase of at least 5% from shoppers who engage with LimeSpot product recommendation boxes. As well, the standard conversion rate often doubles, and can go as high as 5x greater than a regular conversion rate.
Ready to write your personalized product recommendations success story? LimeSpot is a leading product recommendations engine with a patented algorithm, used by thousands of top brands on Shopify and BigCommerce. Whether you're just getting started with product recommendations or looking for a proven solution that will drive more revenue for your business, we can help. Book a demo with us today to share your goals, and we'll help your shoppers experience a more relevant recommendations experience in no time.
