Issue in Product Auto Suggest
Recently a member bought to my attention on an issue related to site search. The frustration was when a customer typed "A3 paper", the auto suggest recommended A3 dividers.
Just to be clear the search results page for “a3 paper” shows the correct search result. The issue highlighted is due to a new feature in FP8 called auto suggest. The reason we are seeing the index divider first in the auto product suggest because of our search relevancy settings.
In the previous versions of WebX, the search results were based on the keywords and product names. This release of WebX the search criteria and weighting of search results is quite advanced.
Below is a table which shows all the key elements in the search relevancy settings in the auto-suggest function. In the case of the Marbig dividers showing first, for the search string “A3 paper “ it is due to the word “A3” is in the category name “A3 Dividers”.
The category name has a higher relevance value as opposed to the product name; this is the reasons why it is showing the dividers first in the auto suggest. To resolve the issues, we will need to reconfigure the relevancy value from category name to name, to ensure the correct item appears.
|Search index field||Default relevancy value|
New Data Search Attributes Filter
On another matter, the data team have completed loading product attributes on the WebX site, see screenshot below. So any search keyword on the site now will show additional enhanced attributes to filter the search results.
There is still work which needs to be completed regarding consistency and fine tuning the attributes, but we now have a base to work with.
For example search on paper can be filtered by CIE, Colour, GSM etc.
There is still work which needs to be done to improve search results for our customers.
Below are the key areas, the technology team will focus on the following tasks in the next two months.
- Fine Tuning the relevancy of search index fields
- Minimum match and phrase slop
- Tuning multiple-word search result relevancy by using minimum match and phrase slop