Ecommerce sites exist to help customers find the products they’re looking for. User-friendly site navigation and high-quality search are the traditional methods for making this happen.
Unfortunately, site navigation can become cumbersome for online shops with a large inventory in many product categories. Likewise, search engines for ecommerce sites are typically not of the same caliber as Google or even Bing, meaning that users need to already know what they’re looking for in order to get relevant search results.
For example, searching for a common phrase such as “jeans” is likely to produce a wide range of results: too many for the customer to wade through.
Website navigation should be convenient for all customers. But is it really possible to create navigation that is useful for everyone? E-shops generally target the center, or rather the ‘average’ shopper. As data reveals, though, this method does not satisfy every customer’s needs.
When customers can’t easily find what they’re looking for, they simply leave the shop and look for the product elsewhere.
The conversion rates of even successful ecommerce shops tend to only be around 2%, with bounce rates being as high as 80%.
Softcube’s customer data shows that visitors to online shops browse on average 5-6 pages per site visit – but the complete catalog of an online shop is typically around 5,000 products! That amounts to 100 pages of products, even assuming a liberal 50 items per page. At a more modest 25 items per page that’s a full 200 pages of products to browse!
In the era of Big Data, it is necessary for ecommerce to find an alternative way to help customers find the products they’re looking for. That alternative is personalization.
The main idea of personalization is to make an educated guess about what a customer will find desirable and be interested in purchasing. This educated guess is made by a computer algorithm and is based on data from a site visitor’s customer journey. Data is collected in real time, and even 2-3 clicks on a website can be enough to begin showing a customer individualized product offers.
Modern personalization revolves around onsite product recommendations. Recommendation blocks on an ecommerce site can take the form of widgets that show items from an online shop’s merchandise. For example, these widgets might display “Customers Who Bought This Also Bought,” or “Products Recommended for You.”
Product recommendation engines, or recommender systems, guess what customers will be interested in by collecting real-time data about site visitors: clicks, items added to cart, purchase history, and more. Placing product recommendation blocks on a product page can increase sales volume by 5-10%. Full site personalization, such as Amazon implements, can generate as much as 30% of site revenue!
The personalized product recommendation blocks discussed above are powered by recommendation engines that interpret large quantities of data (“big data”).
Massive online retailers such as Amazon, Asos, Target, and Tesco can afford to develop their own proprietary personalization services in-house. There are advantages to this approach for large corporations: their data is stored locally; they can develop novel, proprietary algorithms that their competitors don’t have access to; and they can integrate their personalization services at a deep level (direct access to data warehouses and integration with point-of-sale systems in-store, for example).
Deeply integrated in-house data analysis is only profitable for very large enterprises, however. Data scientists and machine learning experts are expensive, not to mention the costs of buying and maintaining mountains of servers and data centers.
But small and medium-sized online stores don’t need to miss out on the action ! Web-based personalization services make it cost-effective and simple for small and medium-sized online shops to offer personalized shopping experiences, integrating recommendations into their existing site.
There are three types of information that product recommendation services care about: information about customers, information about products, and information about the particular e-shop.
1) Customer Information
Customer information includes what customers click on, what products they view, and what they add to the cart and/or purchase – as well as sizes (for clothing), preferred brands, and favorite colors. It also includes information such as the city a customer is located in and their preferred shipping method.
2) Product Information
Product information refers to product titles, descriptions, and images. This information could be gathered from an internal database, but is typically based on the product information that is openly available on a shop’s website.
3) Store Information
Store information includes data about product margins and conversion rates of products as well as shipping terms and other services that the e-shop offers to customers.
These three types of information are all processed by product recommendation services to generate a tailored shopping experience for each and every customer.
There are many bottlenecks in e-shops that can affect the bounce rate of a website. Here are four examples of bottlenecks that personalization can solve and other solutions cannot.
1) Search Results
Personalized search means that a customer’s search results are filtered based on their size, favorite brands, and preferred styles. This ensures that relevant results make it to the first page.
2) 404 Pages
Unfortunately, every website has its occasional error. Instead of losing customers to the 404 page, you can bring customers back into the shop with a well-placed recommendation block.
3) Out-of-Stock Items
Customers are disappointed when the product they want to buy is out of stock. Out-of-stock products should sometimes be removed from a site, especially when there is organic traffic coming in on these pages. Since the items are out of stock, we can’t sell them!
Instead of immediately removing items, however, you can cheer customers up by showing them similar products – similar by color, brand, and other characteristics – to the one that is currently out of stock. Adding recommendation blocks to out-of-stock items can substantially lower the bounce rate from these pages.
4) Thank You Pages
Recommend something else that might be useful alongside everything else that is being purchased. Right before the purchase is finalized is a great time to offer a last-minute suggestion. It’s like that bar of chocolate staring at you in the checkout lane…
We are knee-deep in the age of Big Data. Using Big Data, we can predict customer interests, improve the customer journey, and profit by satisfying customer needs. Failure to act on Big Data insights risks losing not only revenue, but also customer loyalty.