Internal search should help visitors reach relevant content. Treating every entered phrase as a new indexable URL is an additional decision, often made accidentally.
Fictional example: /search/?q=yellow+bag returns three products, and a spelling variation returns the same three. These are two search states, not automatically two editorially planned landing pages.
Establish the scope
Check how search, filtering and pagination construct their URLs. Which addresses are linked internally? Which appear in Search Console? Search phrases can contain personal information, so do not copy raw queries into exports or analytics without review.
For an ordinary internal results list, a noindex instruction may be appropriate. Google needs to retrieve it to read it. Limiting crawling later serves a different purpose and should not inadvertently prevent that retrieval.
A search phrase is not automatically a landing page
If visitors repeatedly seek a clearly defined offering, a maintained category may be a better destination. It needs a dependable scope, suitable results and useful guidance of its own.
After changing the rules, check one search URL with results and one without. Search must continue working. Verify Google’s actual indexing state in Search Console afterwards: seeing noindex on the page does not prove that removal has already happened.
Related check: Distinguish robots.txt from noindex.
An agreed correction to search-page output can be requested through Frontend optimisation, while preserving the search interface.