What is managed hosting?

The word means nothing on its own

“Managed” is not a defined term. There is no standard behind it, no certification, and nothing stopping a provider from applying it to a server where the only thing they manage is the invoice. Two products at the same price with the same word on the box can differ by an order of magnitude in what happens when your site goes down on a Sunday.

That is not necessarily dishonest. A provider genuinely doing operating-system patching and a provider genuinely doing full application support are both doing real work; they are simply doing different amounts of it. The problem is that the marketing page for each says the same word, so the only way to compare them is to ask specific questions and insist on specific answers.

The four things sold as managed

1. Shared hosting with a friendly name

Your site in a directory on a machine full of other people’s sites. The provider runs the machine because they have to — you were never going to touch it. Cheap, fine for a brochure site, and constrained: you get the software versions they offer and the resources your neighbours leave you.

2. A server with managed infrastructure

You get a machine, and the provider looks after the hardware, the network and sometimes the operating system. Anything above that line is yours. This is the most common interpretation and the widest range in practice — “managed” here can mean anything from security patching to nothing at all beyond replacing failed hardware.

3. Application-specific managed hosting

Managed WordPress is the usual example. The provider runs a tuned stack for one application, updates it, caches it and supports it. Excellent if you run exactly that application; useless if you do not, and often restrictive about plugins, cron jobs and database access.

4. Managed service with the server included

The provider is accountable for the site being up, not for a component of it. Sizing, patching, monitoring, backups, restores and migration are all theirs, and the machine is inside the price rather than a separate line. That is the shape of our product, and it is the one where the phrase “managed hosting” does the most work.

Seven questions to ask

Ask these of any provider, in writing, before you pay. A vague answer to any of them is itself the answer.

• What exactly do you patch, and how quickly? Operating system only, or the application stack too? Within hours of a published vulnerability, or when someone gets round to it?
• What happens when an update breaks the site? Do you check after applying, and do you roll back? Who notices first — you or my customers?
• How often are backups taken, how long are they kept, and have you ever restored one? A backup nobody has restored is a rumour, not a backup.
• What is the first-reply time, and does it apply at 3am on a Sunday? “24/7 support” and “a written response time that applies 24/7” are not the same claim.
• Who gets paged when monitoring trips? A human on a rota, or a dashboard nobody is looking at?
• Do you migrate my existing site, and do you rehearse the cutover? Or is “migration support” a document explaining how to do it yourself?
• If I leave, what do I get and how quickly? A full copy of files and database, or an argument?

Our answers to those questions

In the same order, so you can hold us to the same standard as anyone else you are considering.

• Operating system and stack updates are applied and then checked. On Business and above, security patches go on within 24 hours of release.
• We applied it, so we check it and we roll it back. You hear from us in a note rather than from a customer in a complaint.
• Nightly on Essential, kept 14 days; nightly on Business, kept 30; every six hours on Scale, kept 30. On Business and above we rehearse a restore every quarter and send you the result.
• First reply within 4 business hours on Essential. First reply within 1 hour, 24/7 on Business, and first reply within 30 minutes, 24/7 on Scale.
• Uptime checks run every five minutes on Essential and every minute on the tiers above. When one trips, an engineer is paged.
• Migrations are included and counted by the tier — 1 migration, done by us, 3 migrations, done by us, unlimited migrations. We run the site on a private address first and agree a cutover window with you.
• Billing is monthly and you can stop at any renewal. You get a full copy of the files and the database, and help pointing the domain wherever you are going.

Claims that should slow you down

• “Unlimited” anything. Storage, bandwidth and sites are all finite. What the word usually means is “until we decide you are using too much”, which is a policy you cannot read.
• An uptime percentage with no remedy attached. A number in marketing copy is a target. A number in an agreement with a defined credit is a commitment. Ask which one you are being shown.
• Performance figures with no method. IOPS, latency and “x times faster” claims are meaningless without the conditions they were measured under. We do not publish ours precisely because we have not measured them in a way you could reproduce.
• “24/7 support” without a response time. Somebody being awake is not the same as somebody replying inside a window you were promised.

Is it worth it for you?

Managed hosting is worth buying when the operational work would otherwise fall on somebody whose time is better spent elsewhere, or on nobody at all. It is not worth buying because you are nervous about servers in the abstract — nerves are cheaper to fix with a good backup habit than with a monthly invoice.

The honest test: if you already patch without being reminded, know what your last restore actually produced, and would notice an outage before a customer did, buy a self-managed server — from $7.90 a month — and keep the difference. If any of those three is a stretch, the managed plan is not an indulgence; it is the cheaper option once you price a bad weekend properly.

The full arithmetic, with backups added to both sides so the comparison is fair, is on the managed versus self-managed page.