Using AI to Improve Ecommerce Performance

A recent Lucidworks report surveyed retailers to investigate how ecommerce firms with over $100 million in revenue power site performance, including product discoverability, customer experience and personalization. In order to understand buyer intent, a considerable amount of tuning, tweaking, and data combing is required. Retailers are putting relentless pressure on themselves to construct the perfect consumer shopping experience, as they know customers will punish brands that fail to do so. But how does AI fit into the picture?

Research indicated that customers that utilize search capabilities are prepared to make a purchase, but 80 percent of these searches fail due to the site relying on simple keyword search. Providing greater relevancy requires brands to analyze people’s search queries, and then optimizing synonym lists, business rules, ontologies, field weights and countless other aspects of their search configuration.

Given this, what is best practice in e-commerce performance today?

Know Your Customer From First Visit to Final Purchase

Nearly half of retailers (47 percent) said that their add-to-cart (ATC) percentage rate ranges between 11 and 15 percent, while a quarter of retailers (28 percent) average an ATC of 5 to 10 percent. Forty-two percent of retailers reported a clickthrough-rate (CTR) of 16 to 25 percent, with 31 percent of respondents saying their average is above that.

Sixty percent of shoppers visit a website up to four times before making a purchase, with 40 percent making more than five visits before they buy. For the 67 percent of retailers that collect customer feedback signals, each of these visits, in addition to ATC and CTR, provide a better understanding of shopper intent and improve the chances of an upsell or cross-sell.

Incorporate Different Data Sources to Refine Search

Retailers are also leveraging data from multiple sources to form a more comprehensive view of their customers.

76 percent of retailers consider loyalty systems as the most common data source that brands utilize as part of their stack, with point-of-sale (POS) data trailing as a close second (59 percent).

Geographic information system (GIS) technology is tied for third at 40 percent. Without incorporating GIS tech, your website could be promoting a snow shovel to customers in the Canary Islands, or a lawnmower to Mancunians in the dead of winter. Pushing products while being oblivious to the context and behavior of shoppers could lead them to shop elsewhere.

Prepare Site Performance for Peak Demand Periods

Factors including site performance, product findability, and personalized recommendations all combine to impact whether a customer returns later, purchases or completely abandons their search. These factors prove particularly crucial during high-volume shopping periods, like around Christmas. Nearly three-quarters (73 percent) of retailers noted downtime, degraded site performance and poor customer experience, collectively as their top concerns during peak demand periods.

Keeping in mind that 40 percent of potential customers will bounce if a site takes longer than three seconds to load. Ensure site performance, including page-loading speed, can manage the influx of holiday shoppers.

Don’t Let Outdated Product Catalogues Cost You Sales

Modern technology enables retailers to update their product catalogues in less than 30 minutes. Despite this, 53 percent of retailers reported that they require up to 24 hours or more to set up a new catalogue item to be available for sale online. One of the leading causes for this could be the legacy infrastructure, which results in longer lead times, which could cause missed opportunities for more time-sensitive trends.

Adopt AI Tools to Understand Customer Intent

Typically, retailers are not technology experts. However, with recent developments in artificial intelligence (AI) and machine learning (ML), it has become easier to improve site performance, even without being a search expert. Retailers are beginning to adopt AI-powered tools like clustering and classification for ecommerce at a relatively high rate. However, they still have a way to go in order to create hyper-personalized experiences and understand customer intent, as many of these tasks are still being carried out manually. Unsurprisingly, larger retailers with over $400 million in revenue, are leading the pack in adopting AI-based systems.

According to Lucidwork’s report, retailers said that documentation classification, such as assigning items to a specific category to make it easier to find) is currently the most common application of AI and ML (59 percent), with product recommendations behind by a small margin at 56 percent. Natural language processing (NLP) comes in third (52 percent), with retailers working to enhance the customer experience by improving the ability to identify key phrases and terms.

Conclusion

In today’s always-connected shopping environment, factors such as search performance and product discoverability can heavily influence both online and in-store sales. For retailers, the ability to deliver relevant search results, personalized recommendations and real-time responses are what will allow them to pull ahead of the pack in the long run.