Why “Learning Search” Gets A Failing Grade

Learning Search Gets A Failing GradeIn a previous blog, we discussed the concept of “Learning Search” and what the technology can and can’t do. In this installment, we’d like to look a little closer into how it can be used to better effect when incorporated into a system which takes other factors into account as well.

The problem with most search technology which employs learning search is that it is used as the primary sort for delivering search results. The technology simply measures what products matching the search description are clicked on the most and then delivers those products first in the search results. The search results therefore become both a popularity poll and a self-fulfilling prophecy.  (The products that are clicked most get pushed to the top of the page – and the items at the top of a page get clicked most.) And the search function is “learning” only insofar as it is updating its tally of the number of clicks a product is receiving.

Furthermore, when implemented through a sub domain, this activity is taking place away from the retailer’s eCommerce site, degrading SEO in the process.

Learning search can be helpful, however, when taken as part of a larger, more holistic approach to delivering search results. This is where EasyAsk’s approach and technology come in. EasyAsk takes a multifaceted approach to delivering the most accurate, relevant results to a site search inquiry. First and foremost, the cornerstone of EasyAsk’s technology is natural language search software.

By using natural language search, EasyAsk finds the most relevant results because it understands the meaning and intent of the search term – even complex, long-tail search queries. From that point, the delivery and display of the search results can be refined by a formula which takes many factors – including popularity and click-through rate, if desired – into account. This customizable formula, created collaboratively by the retailer and EasyAsk staff, can factor in many variables, including inventory levels, profit margin and sales trends, among others.

The results meet both the customer’s and the merchandiser’s needs and can be adapted on the fly to meet changing market and inventory conditions.  Plus the search and display results are delivered on the etailer’s site, not on a sub-domain page.

EasyAsk eCommerce Edition leads with language relevance and can further sort products in a self-tuning system that takes not only click-through popularity, but also newness, margin, stock level, sales popularity and any other metrics a site owner may decide to include. It’s just a smarter way to go.


Comments are closed

Ready to see how EasyAsk's eCommerce solution can help you? Request a demo!
mp3 database movie database pdf database