Every month, we perform countless site search evaluations for customers, prospects and just about anyone who’d like some advise on improving e-commerce conversions. As part of this evaluation we look at about 30 parameters that help us rate search quality. In this series of posts, I will be talking about four tests that you too can use for evaluating your site search.
Lets start with the first set of parameters.
- Autocomplete Auto-complete is the first thing that visitors have come to expect from an e-commerce search. This is not just a matter of making it look good but has a direct impact on how quickly the visitor can arrive at the product he or she is looking for. By providing suggestions in the drop-down box, we’re shorting the circuitous route he would have had to take by punching in the product name or attributes to search.
- Single-word Spellcheck Visitors make mistakes in entering queries all the time. On some sites, as high as 20% of the search queries can be spell errors. Without spellcheck in the search, at least 20% of the queries will return zero results, causing lost conversion opportunities. Single-word spellcheck test is a simple test where we search with a single mis-spelled word as the query.
- Double-word Spellcheck The aim is the same as the single word spellcheck but in this test, the complexity is increased by introducing spell errors in two words. We’re trying to understand whether the search product’s spell correction engine has enough resilience to detect and correct two words with spell errors in the same query. In some cases, we extend the tests to include triple-word spell errors.
- Synonym Match For a visitor to an e-commerce site, a “cellphone” and a “mobile” are terms used interchangeably. At the core, a search engine is a text comparison system and a “cellphone” query is not the same as a “mobile” query. These searches will end up throwing different search results unless dealt with using a technique based on defining and matching synonyms. Simply put, the product manager must be able to define a relationship that tells the search engine that when a visitor searches for “cellphone”, those products that contain “mobile” are also perfectly valid matches. The search engine is essentially allowing the product manager to say that a “cellphone” is a synonym for “mobile” and vice versa.