Intelligent search: On-site search fine-tunes customer experience
Intelligent search as an integrator
Intelligent search has a very integrative function, according to Stephen Rahal, director of product marketing at Coveo. “Customer data can be unified into a single index by Coveo,” he commented, “so information collected through one channel such as the call center can be accessed in the repair department, and then flow to product engineering.” Integration of Coveo with business applications and interfaces through standard APIs and frameworks allows the data to move across the enterprise. Coveo Analytics can then inform business users about areas of concern, such as signals indicating people are trying to resolve a particular product issue. The issue can be escalated to product management as an alert.
A new user profile service allows Coveo to capture behavioral data, centralize customer data, and show the customer’s past action history, pulling all this data together. This technology enables scaling segmentation in a way that provides unique experiences for each person at each touchpoint. It feeds machine learning in order to respond to queries with smarter results or predict what a person needs through smart recommendations.
Through machine learning and natural language processing, Coveo has developed the capability to provide users with a Q&A search format that produces “smart snippets.” These excerpts present targeted answers, rather than just a list of documents. In addition, Coveo is able to learn in real time from users’ queries. “When users click on a recommendation from options that Coveo provides, the software learns more about the meaning of the words in the search,” observed Rahal, “and it can then provide better product or content recommendations based on the purchase histories of other customers, for example.”
A matter of terminology
One of the challenges in online searching is the tendency of users to describe the products they want using terminology that is different from the terminology used on the website. Search analytics can help discover intent, after which keywords and product descriptions can be adjusted, leading in turn to better search results for the user.
Terminology in search terms was a problem for New Pig, a provider of industrial cleaning and safety products. Supporting a range of industries from manufacturing to medical, New Pig serves more than 300,000 facilities in100-plus countries on all seven continents, offering items such as absorbent mats, wipers, spill containment products, and safe storage containers. Because its products are unique, customers do not always use terminology that matches up with product descriptions and key words.
“We tried outsourcing the search function but it was a black box, and the system did not learn as the vendor predicted,” explained John McQuade, director of software development at New Pig. “We wanted to control the functionality ourselves.” Partly as a result of the terminology issue, customers were getting too many “zero search results” from their queries. Moreover, any changes needed to be modified by IT, while marketing wanted to be able to change keywords. Finally, the company has products that are suitable for COVID-19 response, such as wipers that are compatible with sanitization products and printed floor mats that can be used for directional indicators in store aisles. Being able to onboard these products quickly and make them findable was a priority.