eBay Wildcard Search article from eBay
As we’ve previously announced, eBay is deprecating wildcard queries from search, including web and mobile search, saved searches, and API calls to search. We’ve been working with our Product Team to create documentation around best practices and how you can best move your business forward.
Wildcard queries include an asterisk (*) with their keyword search: The asterisk is intended to match all words starting with the letters in the wildcarded word. For example, the keyword search dress* would include in the search results listings with any of the following in their titles: dress, dresses, dresser, dressers, dressing, dressel, dressage, dressup, etc. in their title in the search results.
Only ~0.35% of eBay queries contain wildcards, but they take a disproportionally high amount of capacity to support. The impact of wildcard queries can have a significant impact on the responsiveness of the site to all user searches, especially at peak times. This impact is why the industry practice, in both web and commerce search industries, is not to support wildcards of this type. Instead, they follow the same approach as eBay in providing automatic query rewriting, which captures the intent of the user query and matches it to items. In addition, advanced search operators such as exclusion and phrasing are provided for users who want finer-grained control over their queries.
Writing effective search queries without wildcards
There are two approaches to searching without wildcards.
The first is to search using a simple keyword search query. eBay is continually improving query processing to make it simpler for users to find what they want. After being processed, a query can include many variants of the original words as well as being expanded through categories or other item-related data. For example, if the user typed gucci handbags as the query, processing could alter that query to search for (gucci or Brand:Gucci) and (handbag or handbags or purse or purses or Category:Handbags). In many cases, searches that used wildcards a year ago can now be run more effectively as keyword queries.
Consider the query anne klein scar*. The intent of this query was to retrieve Anne Klein scarves regardless of whether the seller used scarf, , scarves, or scarfs in the item title. However, the simple keyword query anne klein scarf retrieves all of these items plus items listed in the scarf categories, which may not have the word scarf or scarves in the title. Plus it has the advantage of not including items with Anne Klein and scarlet in the title (e.g., scarlet handbags and jackets): these items would have been returned by the wildcard query. As another example, the query ruffl* curtain* can be replaced by the simple keyword query ruffled curtains and will return all items with ruffle, ruffled, or ruffles in the title that also have the word curtain or curtains in the title or that are in the curtains category.
The second approach to searching without wildcards applies to complex advanced queries. In these cases, the most straightforward way to replace a wildcard is by listing the words separated by commas and enclosed in parentheses. For example, if your query was lego minifig* -lot* you can rewrite it as lego (minifigure, minifigures) -(lot, lots) or as lego (minifigure, minifigures) -lot -lots. This query will match items with the words lego and either minifigure or minifigures but without lot or lots. The parentheses () list alternatives. The minus sign – excludes words.
In general, the advanced search operators (http://pages.ebay.com/help/search/advanced-search.html#using) such as exclusion, option lists, and phrasing can all be used to fine-tune queries. For example, to further constrain the results, a query can contain double quotes ” ” to force the words in the quotes to appear next to each other and in that order. If the user is looking for a filter for a coffee maker, using the phrased query “coffee maker filter” will return only items with those three words in that order in the title; this query will exclude items that are coffee makers with filters.
Note that advanced queries do not undergo any of the automatic expansions that simple keyword queries do. So, the query for coffee maker filters would need to be written as (“coffee maker filter”, “coffee maker filters”) in order to get the rewrite from filter to filters— which would have been included automatically for a simple query. Likewise, the previous advanced search example requires both minifigure and minifigures, as well as -lot and -lots.
Our recommendation to our users is this: First try simple keyword queries, easily combined with category and other constraints such as price. Only if that is not fine-grained enough, move to advanced operator queries.