diff --git a/docs/2-handbook/swarm/archive.md b/docs/2-handbook/swarm/archive.md
index 5019d3a99..8c352208c 100644
--- a/docs/2-handbook/swarm/archive.md
+++ b/docs/2-handbook/swarm/archive.md
@@ -130,9 +130,15 @@ restore fails. Once a week (`deploy.archive.checkSchedule`, default `weekly`)
 the archive host runs `restic check` on every repository. It checks the
 repository's structure in full and reads a random `deploy.archive.checkSubset`
 of its data (default `5%`), so repeated weekly runs cover a repository without
-any one run reading all of it. The check runs on the archive host because the
-repositories are local there, so it reads no data over the network and needs no
-credential beyond the passwords the prune already reads.
+any one run reading all of it. The check runs on the archive host because it
+already holds every repository and the passwords, so it needs no credential
+beyond the ones the prune reads.
+
+With local storage the check reads the repositories from the host's disk.
+With `deploy.archive.backend = "s3"` it reads each repository's structure and
+its subset from the bucket through rclone every week, which costs bandwidth and
+requests on your S3 service. `deploy.archive.checkSubset` is the lever: lower
+it to read less.
 
 The check and the prune share one lock and never run on a repository together;
 whichever starts second waits. A run that fails for an agent logs