% Zero Results
- Record and analyze keyword searches and termless searches that resulted in zero results.
- Hypothesis: As linguistics quality of the search product improves with relevancy tuning and better product data, and as Type Ahead functionality becomes more robust, we should see a decrease in the % of zero results.
- Users should have more habits for browsing/shopping, and we will see a drop in No Matches Found as keyword searches become more familiar to our customers.
- Reason dependent. Some Zero Results are unavoidable.
- Permissions: the user does not have permission to view this product.
- Availability: there is no available inventory for this product.
- No Matches Found
Item Code vs. Keyword vs. Termless (Faceted) Searches
- Map API requests from the user to an analytics dashboard
- Hypothesis: Over time, as the linguistics quality of our search platform improves with relevancy tuning and better data, we should see an increase in keyword searches for users and a decrease in Item Code as shopping and browsing behaviors are more likely to be termless searches / keyword searches.
- Reason dependent. Some level of Item Code specific searches will remain as people build specific shopping carts to meet their needs.
High # of Results
- Record and analyze search terms that create a large number of results
- Hypothesis: As AutoSuggest / Type Ahead functionality becomes more robust with historical searches, descriptions, and popular search – we should see a decrease in the count of users achieving a high # of results.
- We can also consider future features like additional tiers in the taxonomy to drill through for termless search, or creating supplementary pages for products with too many results
- Supplementary pages:
- Pro – provides user opportunity to get to a more specific product faster
- Con – inserts another step in the product, takes them to another view (more of a shopping behavior)
- Reason dependent. Some High # will happen naturally with generic terms, so this KPI should decrease over time.
Image Review Duration
- Record and analyze users interacting with image sliders
- Hypothesis: “What makes a great image?” As image quality improves and images become more used / larger features in the app, we will be able to redesign in a way that makes search more browse-friendly with better product images.
- Measure clicks through images, time spent on images, metrics on average # of image browses per search term
- Trends, times of day / week where people are more or less likely to look through images
Active Usage
% Zero Results
- Record and analyze keyword searches and termless searches that resulted in zero results.
- Hypothesis: As linguistics quality of the search product improves with relevancy tuning and better product data, and as Type Ahead functionality becomes more robust, we should see a decrease in the % of zero results.
- Users should have more habits for browsing/shopping, and we will see a drop in No Matches Found as keyword searches become more familiar to our customers.
- Reason dependent. Some Zero Results are unavoidable.
- Permissions: the user does not have permission to view this product.
- Availability: there is no available inventory for this product.
- No Matches Found
Item Code vs. Keyword vs. Termless (Faceted) Searches
- Map API requests from the user to an analytics dashboard
- Hypothesis: Over time, as the linguistics quality of our search platform improves with relevancy tuning and better data, we should see an increase in keyword searches for users and a decrease in Item Code as shopping and browsing behaviors are more likely to be termless searches / keyword searches.
- Reason dependent. Some level of Item Code specific searches will remain as people build specific shopping carts to meet their needs.
High # of Results
- Record and analyze search terms that create a large number of results
- Hypothesis: As AutoSuggest / Type Ahead functionality becomes more robust with historical searches, descriptions, and popular search – we should see a decrease in the count of users achieving a high # of results.
- We can also consider future features like additional tiers in the taxonomy to drill through for termless search, or creating supplementary pages for products with too many results
- Supplementary pages:
- Pro – provides user opportunity to get to a more specific product faster
- Con – inserts another step in the product, takes them to another view (more of a shopping behavior)
- Reason dependent. Some High # will happen naturally with generic terms, so this KPI should decrease over time.
Image Review Duration
- Record and analyze users interacting with image sliders
- Hypothesis: “What makes a great image?” As image quality improves and images become more used / larger features in the app, we will be able to redesign in a way that makes search more browse-friendly with better product images.
- Measure clicks through images, time spent on images, metrics on average # of image browses per search term
- Trends, times of day / week where people are more or less likely to look through images
Active Usage
- Record and analyze feature clicks and session time
- Additional improvements to this metric can be made with recommendations, reviews, and more
- As we have several well placed engagement points for the users in the browsing experience, as long as they are simple and to the point, we can engage our customers in our catalog for longer stretches of time.