Managed WordPress hosting has gotten genuinely good. Pressable, WP Engine, Kinsta handle the things every site needs — updates, backups, basic security, decent caching, a CDN. That's real value, and worth paying for. It's also not the same thing as having an infrastructure strategy. The decisions that determine whether your site is fast, reliable, and recoverable still belong to you.
There’s a common misreading of what managed WordPress hosting buys you: the assumption that if the host handles the platform, you don’t have to think about infrastructure. The host runs the updates, the host does the backups, the host has a WAF, the host has a CDN. So infrastructure is solved.
This is mostly true for the things managed hosts are good at, and it’s a real value: you genuinely don’t have to think about whether PHP got patched last week. But the parts of infrastructure that determine whether your site is actually fast, actually reliable, and actually recoverable are still yours. The managed host can’t make those decisions for you, because they depend on what your site actually does.
What managed hosting does, and doesn’t.
The good managed hosts (Pressable, WP Engine, Kinsta, Convesio) all handle a similar baseline:
- PHP and WordPress core updates on a sensible cadence
- Plugin auto-updates if you opt in
- Nightly off-site backups
- A basic Web Application Firewall
- An edge CDN with sensible defaults
- Server-level caching for non-logged-in traffic
- SSH and WP-CLI access
- Staging environments
This is genuinely the table-stakes layer. You shouldn’t be running WordPress without it in 2023. But it’s the floor, not the ceiling. Three categories the host can’t make decisions about:
Performance: the host can serve fast bytes. The site has to send the right bytes.
Managed caching is fast at serving whatever the site rendered. It can’t help you with a WordPress install that ships 2 MB of CSS and 1.5 MB of JS on every page load because of an off-the-shelf theme. The host’s CDN delivers that 3.5 MB efficiently. The visitor still waits for 3.5 MB.
Performance decisions that belong to you:
- Whether your theme ships only the CSS and JS each page actually needs (versus a giant bundle)
- Whether image uploads are sized and compressed appropriately for their use
- Whether your data model leads to N+1 query patterns in templates
- Whether your
wp_optionsautoload table has crept into the megabytes - Whether your third-party scripts (analytics, embeds, chat widgets) are blocking render
None of these is the host’s problem. All of them are why a site on the same host can be twice as fast as a different site on the same host.
Reliability: hosts handle uptime. They don’t handle your dependencies.
Managed hosts have strong uptime numbers. The thing that takes a WordPress site down is rarely the host. It’s usually:
- A plugin update that conflicts with another plugin
- A PHP version bump that surfaces deprecated syntax in old custom code
- A third-party API the site depends on going down
- A WordPress version that drops a feature your theme relied on
- A database backup that wasn’t actually being tested for restore
The host’s 99.99% uptime guarantee doesn’t catch any of those. The decisions that catch them — what to test, what to monitor, what to stage before deploying, what to keep in source control — are infrastructure choices the host doesn’t make.
Recoverability: backups exist. Restores have to work.
Every managed host backs up nightly. Almost none of them require you to ever test a restore. Which means a backup is a hypothetical until the day something breaks and you discover the restore process doesn’t actually produce a working site because of a missing file, a misconfigured environment variable, an extension that didn’t come back.
Recoverability work that belongs to you:
- Periodically test a restore to a staging environment and verify the site works
- Document the restore procedure so a future developer (or panicked stakeholder) can execute it
- Identify what’s NOT in the host’s backup (uploads off-site, third-party content, environment-specific config) and have a separate plan for those
- Define recovery time objective (how fast can the site be back) and recovery point objective (how much data are you willing to lose) and verify your setup meets them
These are infrastructure decisions the host facilitates but doesn’t make.
What it actually means to think infrastructurally.
The shorthand: managed hosting buys you the floor. Infrastructure thinking is everything that determines how high your ceiling goes. The two compound: a good infrastructure strategy on bad hosting is a fragile thing, and good hosting under no strategy plateaus quickly. Both matter.
The work that platform architecture handles is the strategy half: making the choices the host can’t make for you, and documenting them so the next person inheriting the platform doesn’t have to rediscover them. Hosting is a checkout. Infrastructure is a discipline.