Understanding how to meet a question with the right answer isn’t a linear process, however—it’s one that has stages and variables at every stage. In other words, there’s a lot of chances for error between hitting “search” and seeing results. That’s why semantic search is so fundamentally powerful: it works to understand the question, to better provide the answer.
Understanding the question
- “Do you have any striped blue and yellow silk ties available for purchase?” This is the type of question that your ecommerce site may field if a customer calls in or chats with you
- “Blue and yellow tie.” This is the same question as the one above, typed into your on-site search box
The two questions above may be asked by the same person, but depending on the medium they’re being asked within, they’re obviously taking a different form. Unfortunately, the search box pushes people to be more succinct, which is why there are critical details missing from the shorter second question
Having a semantic search function that translates the second question into the first to comprehend it means being able to provide that customer with the same degree of results that a real, live person would be able to. Take a look at how semantic search works to understand the question as it was meant to be asked:
- “Blue and yellow” indicates that the customer is searching for a tie that has both of these color properties. All ties that don’t have blue and yellow in them will be eliminated from search results
- “Blue and yellow” together in the search query also means that the search will need to be further refined by pattern. But, as the customer did not provide design or pattern options, these refinements will be displayed as options on the search page, allowing the customer to manually check the box for “striped” if they don’t immediately see their result
- Because no material was designated, semantic search will also assume that the customer has a preference that has not been indicated, giving them a “material” refinement section on the results page
- “Tie” tells the search function that the customer only wants ties, not a “bowtie” or “cravat” or anything else. Tie is a specifier and a hard trait that will be semantically understood as central to the query
Using logic and Natural Language Processing (NLP) technology, semantic search will turn “blue and yellow tie” into the question “can I see any neckties you offer with both blue and yellow colors in them, with the ability to pick from any materials or patterns you might have?” This is a question your customer might not outwardly ask, but it’s encompassing of their needs!
Answering the question
Once the question has been understood, a semantic search function will work to answer in the best, most applicable way possible. For your ecommerce site, this is a valuable asset in place of a salesperson: NLP will work to understand the customer without going back and forth to get more information, presenting options where questions might otherwise have been asked.
At the end of the day, every search query is a question and every question needs an answer. Semantic search is going to provide the right answer—even when the question is ill-defined or wholly unasked! You can’t ask for a more intuitive sales channel.
About Celebros
Celebros - Magento’s Premier Technology Partner for Site Search – ECG approved for both Magento 2 and 1.X
Celebros is the global leader in Natural Language Site Search, Merchandising and Navigation technologies for online retailers. Through intelligent concept-based and semantic algorithms, Celebros’ Site Search understands the shopper’s intent, queries and personalized search terms. In return, Celebros’ natural language Site Search delivers relevant and accurate results the first time and every time on the first page resulting in higher conversion rates, increased AOV’s and turning more visitors into paying customers.
Magento merchants receive a feature rich suite of conversion boosting tools that include:
- Natural Language Site Search – Machine Learning
- Advanced Auto Complete
- Guided Navigation
- Faceted / Multi Select Navigation – Machine Learning
- Product Recommendations
- Product Finder / Guided Shopping
- SeeVolution Heat-mapping Analytics
- Dynamic Merchandising Engine
- Virtual Categories / Dynamic Landing Pages
