Cache hit ratio explained, and why 85% is a network average

Hit ratio is hits divided by requests. How to raise it with TTLs, cache keys and purge, and what Aptranet 85% average actually means.

Get started Performance

Hit ratio is the share of requests the edge answered without origin. Aptranet publishes an 85% average across the network. Your site will match that only if the bytes are actually reusable and the keys are not poisoned.

What counts as a hit

A cached object served from a PoP. Misses include first views of new URLs, expired TTLs, purges and anything you bypassed.

How teams accidentally tank it

Varying on every cookie, putting unique query strings on static files, setting no-store on HTML, and purging the whole distribution on every save.

How to raise it

Long TTL on hashes, ignore irrelevant query strings, cache anonymous HTML, bypass only stateful paths, purge narrowly, prefetch known launch URLs.

Averages are not your dashboard

85% is the network. A tiny site with unique HTML every view will sit lower. A software download portal should sit higher. Measure per path.

Continue on Aptranet

Frequently Asked Questions

Is a higher hit ratio always better?

Not if you are caching personalised HTML to get the number up. Correctness first.

Does Origin Shield change hit ratio?

It changes how misses reach origin, not whether the edge had the object. Both matter.

Should I chase 99%?

Only if your workload is almost all immutable objects. HTML-heavy sites will not, and should not, sit there.