That said, I have heard this often enough that some people must do it. I can't imagine why though, as it takes only a few seconds for a human to scan a resume. Even at 10 seconds per resume (a conservative estimate), that means the human could screen 360 resumes per hour.
Based on experience at firms with on the order of 1:1 applicants to hires up to 10,000:1 applicants to hires, I’d agree most such tools are not present at most companies, as they tend to be used from ratios of 100:1 and up.
On the other hand, while most companies might not have or use such tools, the more jobs a company has, the more likely they are to have some form of selection or filtering in the pipeline.
But none of the ATSs really collect great data on how many of their customers use them in their core processes. From observing actual, heavy duty recruiter and HR usage at Ladders, I suspect it is not very many.
These are people people after all, and they are not great at writing REGEX style queries to find only the precisely correct resumes they are looking for. You also need to remember that they are frequently working on novel searches each week, and wouldn’t have the confidence or the capability to craft such searches.
Tech companies doing recruiting of any meaningful volume are indeed generally using searches or matches of some kind at the start of the funnel.
I'm sure this happened somewhere at some point in time, but IMO it's becoming a myth blown out of proportion.
I've never actually worked at a tech company that had any sort of automatic resume filtering software. As a hiring manager, we didn't even have HR pre-screening resumes for us. If we did have recruiters, they were competent and incentivized to find good candidates (e.g. wouldn't be unnecessarily rejecting good candidates because they were missing exact spellings of keywords)
It would seem ludicrous in 2022 and this hiring crunch for a tech company (one that you'd actually want to work for) to arbitrarily filter out resumes based on incomplete keyword searching.