This feature would entail adding another search operator to the existing set which enables users to search for a pair of keywords within a certain number of words or characters of each other.
Proximity searches are useful when searching through catalog-style websites and in other contexts where descriptive words are important to the relevance of a result but may not appear in a particular order that can be searched for verbatim in relation to a broader term.
The addition of this feature should have no impact on existing workflows or UX.
One of the most vexing issues with verbatim searches is the problem of phrasing. You may want only results for, say, copper pipe, but sites selling copper pipe may not use that phrasing. Your search for copper + pipe will yield results for, say, iron pipe that mentions copper elsewhere. A search for "copper pipe" excludes relevant results phrased as "Pipe, Round (200mm), Copper" or similar.
While result rankings do a reasonably good job of implicitly ranking results by proximity, the ability to explicitly limit results by proximity is one of those capabilities that's rarely needed but overwhelmingly helpful when necessary.
Many search tools (historically including Yandex, Bing, and Google) include a proximity search operator to facilitate explicit proximity searches. By specifying a maximum number of words or characters between search terms, the user is able to search for precisely what they need while still accommodating differences in phrasing. Our example search for copper pipe could be specified as copper NEAR(4) pipe to yield results containing both terms within four words of each other. This would successfully exclude the first case and include the second.
English documentation for the three search engines noted above no longer indicates the availability of this feature, and attempts to perform searches using the formerly-available operators did not yield expected results. This feature is still widely available in line-of-business (eDiscovery, library indices, etc) and other smaller-scale search software.
The implementation I am most familiar with (Microsoft Purview) documents this feature as follows:
NEAR (keyword1 NEAR(n) keyword2): Returns items with words that are near each other. In the keyword1 NEAR(n) keyword2 syntax, n equals the number of words exclusive of keyword1 and keyword2. For example, to identify instances where the term best is within 3 words of worst (example sentence, 'Best is opposite of worst.'), you would use best NEAR(3) worst. This returns any items where there are 3 or fewer words between best (keyword1) and worst (keyword2). If no number is specified, the default n value is 8.
(https://learn.microsoft.com/en-us/purview/ediscovery-keyword-queries-and-search-conditions)
