There are 6 quick and easy ways to measure the quality of your on-site product search. If any of these metrics are performing below your expectations, there is likely room to improve. Even a small increase in search results quality can have a huge impact on conversion rates and your bottom line.
1. Site-Search Conversion Rate Compared to Visits Without Search and website average
No matter how poor your search is, visitors with search will generally convert at 2x the website average conversion rate and 4 -13x versus visits without search. This is because searching visitors typically have more specific shopping intent than visitors who are simply browsing.
2. Search Exit Rate
A good rule of thumb to judge if your site-search needs improvement is if the exit-rate on search result pages is close to or on par with your website's average bounce rate. Your search result pages should have some of the lowest exit rates on your site, since those searching have intent and often only exit search pages if results are irrelevant or no results were actually presented.
3. Your Null-Result Rate
(i.e. the percentage of total searches executed that return zero results). If that rate is above 5%, there is likely room for improvement (how to measure null-result rate).
4. Test Your Search for Accuracy
Search for words (singular, plural, misspelled) and see if the same results appear. For example, if you're in sporting goods, a search for 'basketball', 'basketballs' and 'bascetblls' should all return the same results. (how site-search spellcheck works)
If you are promoting certain content/products and have applied some sort of “boost” rule, make sure it doesn’t override the general relevancy of the results. Do a search for something totally unrelated to the content/product that is being “boosted” but that contains one word that is found in the title of the boosted product/content. If that product/content comes up in the results despite being totally irrelevant to the actual query, then you know you are also having trouble combining business rules with basic text-matching search. (boost and bury rules explained)