“Which hosting is best?” sounds like a buying question, but it hides the information that decides the answer. A brochure site, a busy publication and an online store can all run WordPress while carrying very different consequences when an update fails at 2 a.m.
Consider three owners. A photographer changes a portfolio twice a month and would rather call support than learn server administration. A local clinic relies on appointment requests and cannot afford a silent contact form. A small software team needs a background worker and a custom service that ordinary WordPress plans do not allow. Giving all three the same recommendation would be convenient and wrong.
This guide compares shared hosting, managed WordPress, managed VPS or cloud hosting, and a self-managed VPS. Provider labels are inconsistent, so the name of a plan is never accepted as proof. The decision comes from five operational scores and one written promise: who does what when the site breaks?
Start with a Workload Card
Before opening a pricing page, describe one normal week and one bad day:
- What software must run: WordPress only, an online store, scheduled jobs, or another application?
- How often does content, inventory or customer data change?
- How many people publish, administer or develop the site?
- What stops when the site or email fails?
- Who can investigate an incident, and during which hours?
- How quickly must a usable copy return after failure?
Do not invent a precise visitor number if the site has not launched. Record a range and the events that could create a spike: a campaign, product release, news mention or seasonal sale. The card should describe work, not justify a plan someone already wants to buy.
Score five operational factors from 0 to 3
| Factor | 0 | 1 | 2 | 3 |
|---|---|---|---|---|
| Change rate | Rare static changes | Weekly publishing | Daily content or orders | Continuous business data |
| Downtime impact | Minor inconvenience | Lost enquiries | Lost transactions or operations | Safety, contractual or major revenue impact |
| Traffic variability | Small and predictable | Occasional campaigns | Regular large spikes | Unpredictable high demand |
| Custom stack need | Standard WordPress | Common plugins and jobs | Special services or configuration | Multiple applications or deep server control |
| Operations capacity | No server operator | Developer available occasionally | Named operator with monitoring | Team with on-call process and tested recovery |
Do not simply add the numbers and buy the most expensive result. Two patterns matter more than the total. High downtime impact with low operations capacity points toward a service with stronger management and support. High custom-stack need with high operations capacity makes server control more valuable. A self-managed VPS does not become suitable because the website is important; importance increases the need for competent operations.
What each hosting category usually means
Shared hosting
Multiple customer accounts use a provider-managed server platform. The provider operates the underlying hardware and common services, while the customer manages the site within account limits. It can fit a simple portfolio, early brochure site or low-risk project when the provider offers usable backups, current software and responsive support.
Check resource and file-count limits, backup retention, restore charges, email limits and what happens during traffic spikes. “Unlimited” should lead to the acceptable-use policy, not end the investigation.
Managed WordPress hosting
The service is designed around WordPress and may handle some combination of installation, caching, updates, backups, staging, monitoring and specialist support. WordPress’s own hosting documentation notes that WordPress-specific services may include backups, updates or developer tools. The word managed is not a standard scope, so ask which layers are actually managed.
This category often fits a business that depends on WordPress but does not employ a server operator. It does not remove the owner’s responsibilities for access control, plugin choices, content, vendor accounts and independent recovery evidence.
Managed VPS or managed cloud
The customer receives more isolated resources or a cloud instance while a provider or operations partner manages an agreed portion of the system. It can fit a growing publication, store or custom workload that needs more control than a shared plan but lacks an internal team for operating the entire stack.
Read the management boundary closely. Some services patch only the operating system. Others include the web server, database, monitoring and incident response. Write excluded work beside the monthly price.
Self-managed VPS
The customer controls the server and accepts responsibility for hardening, patches, firewall rules, backups, monitoring, capacity, incidents and recovery. It can be appropriate for a team that needs a custom stack and already has the skills and time to operate it. Root access is a capability, not a maintenance service.
A cheap VPS can become the most expensive choice when nobody owns the weekend alert or restore procedure. Count operator time, security maintenance and offsite backups in the real cost.
Apply the score to four real website journeys
1. Portfolio or freelancer site
The site changes monthly, collects a few enquiries and has no custom service. Operations capacity is zero. Shared hosting can be reasonable if restore is simple, support is reachable and renewal cost is acceptable. Managed WordPress may be worth the difference when the owner values staging, automated updates and specialist help. A self-managed VPS adds responsibility without solving a business requirement.
2. Local business lead site
The traffic may be modest, but forms, phone details and local search visibility produce revenue. Downtime impact is higher than the visitor count suggests. A well-supported shared service may still fit, but the buyer should test form delivery, backup restore and support response. Managed WordPress becomes attractive when nobody on staff can diagnose a failed update.
3. Growing content or affiliate publication
New content arrives frequently, campaigns create spikes and the archive becomes valuable. The useful questions are how caching behaves, how resource limits are enforced, whether staging mirrors production and how quickly a full database-plus-media restore can be completed. Managed WordPress or a managed VPS may fit as the workload grows. Move because measured constraints appear, not because a page-view myth says every publication needs a VPS.
4. E-commerce or operational site
Orders and customer records change continuously, so a nightly backup may still lose a business day. Downtime, checkout failure and slow recovery have direct cost. The plan needs a documented recovery point, recovery time, monitoring route and support escalation. A managed platform can reduce the operations gap; a capable team may choose managed or self-managed infrastructure when custom requirements justify it.
Use five decision gates before comparing providers
- Software gate: Is the workload standard WordPress, or does it require other services and server-level configuration?
- Operator gate: Who receives an alert, investigates it and has permission to fix the system?
- Recovery gate: How much recent data can be lost, and how long can the site remain unavailable?
- Change gate: Can updates be tested in staging and reversed without rebuilding the site?
- Exit gate: Can the owner obtain files, database, DNS records and a current backup without the provider’s cooperation?
If any answer is “we do not know,” price the unknown as a risk. A plan is not ready for purchase until the responsible person and recovery route are named.
Check the technical floor, then the operational ceiling
For a WordPress site, compare the provider with the current official WordPress requirements. As of September 2026, that page recommends PHP 8.3 or greater, MariaDB 10.11 or MySQL 8.0 or greater, and HTTPS. Requirements change, so check the source again when purchasing.
The technical floor only shows that WordPress can run. It does not prove the plan can recover your site. WordPress’s backup guidance explains that a typical recovery needs both the database and site files, and recommends keeping several recent copies in different locations. Ask for the retention period, backup location, restore method and last successful restore test. A backup button without a verified restore is an assumption.
Build the provider shortlist
| Question | Evidence to request |
|---|---|
| What is managed? | A written list of included and excluded layers |
| What triggers throttling? | Published CPU, memory, process, bandwidth and file limits |
| How are backups restored? | Retention, location, restore steps and any fee |
| Who responds to incidents? | Support hours, channel and escalation path |
| Can we test changes? | Staging behavior and deployment workflow |
| Can we leave? | File/database export and cancellation terms |
| What will year two cost? | Standard renewal price and required add-ons |
Shortlist providers only after the hosting category and required evidence are clear. This keeps affiliate commission, introductory discounts and familiar logos from deciding the architecture.
The decision record
Save the Workload Card, five scores, chosen category, rejected alternatives, management boundary, recovery target, renewal price and owner. Review it after the first 30 days and again before renewal. If real traffic, support experience or maintenance work differs from the assumptions, change the decision record before changing the server.
The best hosting is not the plan with the longest feature list. It is the service whose responsibilities match the team’s capacity and whose failure can be recovered within the business’s tolerance.
How we researched this guide
This guide uses current WordPress requirements, hosting documentation, server-administration guidance and backup guidance. It is provider-neutral and contains no affiliate links. Hosting scopes and limits vary; verify the provider’s current contract and technical documentation before purchase.
