Intelligent search: On-site search fine-tunes customer experience

Relevant content

New Pig selected Lucidworks’ Fusion as its search solution, and experienced improvement on multiple levels. “Marketing can now adjust key words to make them more compatible with the intent of users’ searches, and then make suggestions to the creative department, which manages copy changes,” McQuade noted. In addition, New Pig is using Fusion Analytics along with Google Analytics to feed purchase information back into the system. “We can see results becoming more relevant based on the wisdom of the crowd and user popularity,” he added. “We also can see if underperforming queries have improved based on adjustments we have made.”

Smarter personalized recommendations are expected as New Pig expands its use of Fusion AI and signals feedback. “We see an improvement already in recommendations and can adjust the product mix presented in the searcher’s results based on the customer’s previous history,” said McQuade. New Pig also has a considerable amount of information about safety, such as how to identify possible sources of floor hazards (leaky roofs and pipe condensation), as well as regulatory information. This information is expected to be increasingly utilized as New Pig continues to leverage Fusion to help customers locate content relevant to their issues, said McQuade.

The Lucidworks Fusion Platform offers search and data discovery using a cloud-native architecture. Its capabilities include support for Python machine learning models and Predictive Merchandiser, which optimizes customers’ shopping experiences. “Our philosophy is that our search solution needs to capture intent, make sense of the user’s request, and apply it to the results,” said Vivek Sriram, chief product officer at Lucidworks. “These concepts support the fundamental premise of our product. Workflows can be made better when user intent is understood.”

The primary use of AI in Fusion is for machine learning. “We are helping users look for patterns, such as logs of product defects, or suggestions about products,” Sriram continued. “In R&D, researchers might want to find others working on similar topics.” Natural language processing adds relevancy by revealing meaning beyond keywords or statistical text analytics. “Search results should be highly contextual,” Sriram stated, “and should depend on many factors, including the role of the searcher, geographical location, and other factors, to maximize relevancy.”

Tuned to a vertical

Some search software solutions are designed to cover a broad range of industries and application types and need to be tuned to the industry as part of the implementation process. Another approach is to design the search solution for a specific vertical so that it is already tailored to usage in that industry. Funnelback software was developed by a small Australian-based research organization to search government and university websites containing extensive text information that was largely unstructured. The company was spun off as a separate entity in 2005.

“Universities and government agencies often have rather small teams managing their websites,” said Jesse Swingle, director of marketing at Funnelback. “Their needs are quite specific, and it helps to have a product that is ready for their environment right out of the box.” For example example, people visiting a university website usually fall into one of a few categories: a prospective student, a faculty member, or a researcher who wants to learn something about the university. “The student might want to know if the university offers a program in a certain discipline,” said Swingle, “while the faculty member might want to find a document that explained a university policy.”

Community Colleges of Spokane (CCS) provides a wide range of traditional educational opportunities and specialized services such as rural outreach and literacy to over 30,000 students per year in Eastern Washington. As part of an overall revamping of its websites, CCS wanted to replace its legacy search appliance with a cloud-based search engine that had greater functionality and flexibility.

Accessibility of content

After some demos and comparative analyses, CCS identified Funnelback as a good candidate and suggested to the consulting company that was assisting with the website redesign that the product could be a good fit for its search needs. “The competitive price and rich feature set, including accessibility auditing, were big factors,” said Gretje Witt, web designer for CCS. “The consultants agreed and also saw the potential for Funnelback to solve some of our multi-use content brokering needs.”

A strong search solution was content on the CCS websites. “We have so many topics and calls to action,” continued Witt, “it would be very difficult for users to find what they need just through navigation.” Users’ ability to locate content has greatly improved since Funnelback was deployed, and feedback has been positive. The use of analytics helps fill the gaps if searches are not producing results. “In some cases, we offer prompts that help guide the user,” added Witt, “while in other cases, we discover that additional content is needed, and we create it.”

Accessibility is very important in educational environments, and this component of Funnelback has been very helpful. “We need to ensure accessibility compliance, so it was a huge bonus that Funnelback came with an accessibility auditing tool,” Witt commented. For the future, Witt plans to continue customizing Funnelback to fine-tune its performance. “We think we have just scratched the surface of what we can do with Funnelback,” she said. “As we get to know the product, we will be able to tap into its capabilities and improve our search performance even more.”