The retail landscape is in a difficult place at the moment. Between economic stressors like high inflation rates, talent shortages and industry changes such as rapidly evolving customer expectations and surging demands, many retailers are scrambling to find a foothold in these uncertain times. They need to make every resource count.
Merchandising teams are critical to e-commerce success but are being dragged down by old tech and outdated methods. According to research, most merchandising teams spend more than half of their time curating and managing manual search rules. This is all thanks to rule-based merchandising, which was developed to make up for the shortcomings of legacy search solutions.
The Shortcomings of Rule-Based E-Commerce Merchandising
Most search solutions are built on top of keyword matching technology, which takes a user’s search query and runs it against the site’s product catalog in looking for products that match. This wouldn’t be a problem, except that the human language has a lot of variation. For example, one person may search for navy tennis shoes while another searches for dark blue sneakers. As humans, we know both queries point to the same product, but keyword matching views them as separate things.
To account for this, merchandisers implement manual rules or synonyms — e.g., navy is the same as dark blue — to ensure the engine delivers search results relevant to the query. However, as traffic grows to a site and language continues to evolve, retailers have to implement more and more search rules to keep up. A large enterprise can easily have tens of thousands of merchandising rules, all of which need constant maintenance and updates because these rules are static. If a merchandiser pins an item to the top of the search results and then they run out of stock, that item will stay pinned, ultimately costing the retailer sales and negatively impacting the customer experience.
Over the years, dozens of features have been introduced to improve or manage the inefficiencies of keyword matching technology. All these have done is make it easier to manage merchandising rules. They don’t address the core problem — the need for such rules in the first place.
A New Approach: Revenue-Focused Merchandising
Recent breakthroughs in artificial intelligence and machine learning (ML) make it possible for retailers to change their approach to search using AI-first solutions. Unlike other AI search applications, which build on top of legacy tech, AI-first solutions have rebuilt search from the ground up with an AI-first approach. By leveraging the abilities of AI at a foundational level, these solutions address all challenges of modern search without the drawbacks of rule-based merchandising.
For example, traditional e-commerce merchandising required thousands of manually created and curated search rules in order to compensate for the shortcomings of keyword matching technology. However, even with thousands of rules, most merchandising teams can only optimize the top 5 percent of searches for an enterprise-sized e-commerce site. This means 95 percent of searches aren’t curated, and curating rules for every possible search is simply beyond human capabilities. So why are we still doing this manually?
AI-first solutions can dynamically manage, optimize and curate results for all searches because they’re self-learning and autonomous. The solutions learn from input data such as user behavior, so they improve with every search. This enables retailers to go from having thousands of search rules on a legacy solution to having just a handful (and often only one or two) for an AI-first solution.
In addition to optimizing every search while dramatically reducing search rules, AI-first solutions also excel at inventory management. Because of their dynamic, self-learning nature, AI-first solutions can unhide inventory that shouldn’t have been hidden to start with. They can boost hidden products that match current trends, and bury those products once they run out of stock. This dynamic reordering helps optimize every search and drive site-wide revenue improvement.
With the burden of rule-based search removed from their plates, merchandisers gain hours back in their day. Instead of spending their time adjusting search rules or programming in new synonyms, merchandisers are now free to focus on revenue-generating initiatives. Activities like A/B testing, curating campaigns and developing sophisticated inventory management strategies, which are already part of a merchandiser’s expected duties, can have more time devoted to them.
But it’s the neglected initiatives that really shine. “Nice to have” initiatives that teams simply didn’t have time for before, such as investigating new technologies, speaking to customers, developing inventive ways to include reviews and user-generated content, and fixing process and data problems that impact site performance, are all activities merchandisers could take on now that maintaining search rules is off of their plates.
The Future of Merchandising
A recent survey found that 25 percent of all workers fear they will lose their jobs to AI, and that number almost doubles when looking at industries like advertising and logistics. However, AI-first solutions aren’t here to replace merchandisers — and truthfully they’re not ready to. They still require human oversight. However, by almost completely eliminating the need for search rules, AI-first search and product discovery is game changing for retailers. Merchandisers no longer have to spend their time fixing the shortcomings of their search engines. Instead, they can focus on more strategic areas, driving revenue for the business and become the strategic partners and domain experts merchandisers should always have been.
Roland Gossage is the CEO of GroupBy, a product discovery platform powered by Google Cloud Retail AI.
View Original Article
