Comparisons

Mac file search speed, measured

Updated 2026-08-05 · 6 min read

Every file finder claims to be instant. Here are real numbers from a real disk, and how to read the ones you see quoted.

The numbers on our own index

Measured on a development Mac holding 6,171,843 files, through the shipping search path, not a microbenchmark:

Query	Time	Matches
roadmap	1.4 ms	76
wooqer	1.7 ms	1,404
invoice	3.1 ms	199
screenshot	3.4 ms	1,464
report	13.3 ms	9,171
clock	13.7 ms	7,463

Two things worth noticing.

Common words are slower. clock takes ten times roadmap, because the trigrams that make up clo and ock appear all over a developer's disk, so far fewer blocks of the index can be skipped. Any indexed finder has this property; a single quoted number hides it.

Building the index costs too. That trigram index takes 1.2 seconds to build over 6.1M files and 24 ms to load from cache on later launches. The initial full scan is 147 seconds, about 42,000 files per second.

How to read a vendor's number

Cling advertises results in under 100 ms on a typical search across nine million or more files. That is an upper bound over a larger index than ours — not a like-for-like against the table above, and we are not going to pretend it is. The only comparison worth anything is both tools on the same disk with the same queries.

Three questions to ask of any published figure:

1. How many files? A number without a file count means nothing.
2. Which query? A rare word flatters any index. Ask about a common one.
3. Does it include drawing the results? Matching is milliseconds; resolving paths and

icons for the visible rows is often the larger cost.

Does any of it matter

Below about 50 ms, no. Both Cling and Everywhere are far under human perception, and the difference between 2 ms and 30 ms is invisible while you are still typing.

What you *do* notice: whether results update per keystroke or per return press; whether the first search after launch is slow; whether the app stalls when a drive is plugged in. Those are architecture, not speed, and no benchmark reports them.

The honest position

Speed is table stakes in this category and stopped being a differentiator once both tools went below 100 ms. Coverage — which files are in the index at all — is what actually decides whether you find the thing.

Common questions

How fast is indexed file search on a Mac?

On a 6.1 million file index, measured queries ran from 1.4 ms for a distinctive word to 13.7 ms for a common one, through the full search path.

Why are some searches slower than others?

Common letter sequences appear in most parts of the index, so far less of it can be skipped. Rare words eliminate almost everything immediately.

Is a faster search app worth paying for?

Below roughly 50 ms the difference is imperceptible. Coverage — whether the file is in the index at all — matters far more than milliseconds.