"Managed" is not a defined term. It means whatever the host has decided it means, and the gap between what customers assume and what is actually covered causes most of the trouble.
Assuming backups and security are handled
Some managed hosts take daily backups with 30-day retention and will restore for you. Others take one weekly snapshot, keep it for seven days, and consider a restore a paid service. Both call themselves managed.
Ask three concrete questions before signing: how often, retained how long, and who performs a restore. Vague answers here tend to predict vague answers later.
Installing a caching plugin on top of server caching
The host already runs page caching at the server or CDN layer. A well-meaning developer installs a caching plugin as well, and the two now disagree about what is stale. The usual symptom is a change that appears for some visitors and not others, or a logged-in user seeing a cached page meant for anonymous visitors.
Find out what the host already caches before adding anything. Many managed hosts publish a list of plugins they block precisely because of this.
Plugin sprawl
Thirty-plus plugins, several doing overlapping jobs, a few unmaintained for two years. Each one is code running on every request and a potential vulnerability, and the abandoned ones are how most WordPress sites are actually compromised — not through a clever attack, but through a known flaw in a plugin nobody updated.
Audit quarterly. Anything not updated in a year needs a justification to stay.
Staging provided but never used
Most managed plans include one-click staging. Most customers still edit the live site, because staging means remembering to push changes back. Then a plugin update breaks the checkout on a Friday evening.
The discipline matters more than the tooling: update on staging, look at the pages that matter, then push. It takes ten minutes and removes the entire category of "the site broke after an update".
Ignoring the PHP version
Sites run for years on a PHP version that stopped receiving security fixes. It keeps working, so nobody touches it — and the longer it is left, the more likely a jump will break an old plugin, which makes everyone more reluctant.
Upgrade on staging, one major version at a time. The pain grows with the delay.
Database bloat
Post revisions accumulating without limit, expired transients never cleaned, orphaned metadata from plugins removed years ago. A database that should be 50 MB reaches several hundred, and every query gets slower.
Cap revisions, clean transients on a schedule, and check table sizes occasionally.
The question worth asking any managed host
"My site is down at 9pm on a Saturday. What happens?" The answer tells you more than any feature list — whether anyone is on duty, what the response commitment actually is, and whether restoring a backup is something they do or something they walk you through.
About the author. Kislay Chandan has worked with schools since 2014. Rather than gathering requirements from the outside, he worked inside a school to see what actually goes wrong day to day, and built what he found into the software. He still works with Shemrock Glitters in Patna — most changes to the product start with something that came up there.
Gallery
