This guide is for business owners and technical decision-makers choosing where to host a website or web application, and trying to navigate the gap between the headline price and the actual cost. By the end you will know the differences between shared, VPS, dedicated, and managed cloud hosting in practical terms, when each one is the right answer, the line items that usually get missed, and what changes when you move from one to the other.
Who This Guide Is For
Founders, operations leads, and technical managers responsible for a business website, e-commerce store, or web application — anything from a brochure WordPress site to a custom Laravel platform. You are either starting fresh and choosing where to host, or sitting on shared hosting that has started to creak and wondering whether the upgrade is worth it.
Before You Start
You should know roughly what you are hosting (a content site? a SaaS app? a store?), the rough traffic levels, and whether the site processes sensitive data — payments, personal information, anything regulated. The right hosting tier depends on those answers more than on price.
If the question is more architectural than infrastructural — for example, whether to host a WordPress site or build a custom platform — How to Choose Between WordPress and Custom-Built is the right starting point.
The Four Categories That Actually Matter
Hosting marketing language obscures the differences between options. The useful categorisation is:
- Shared hosting: many websites on the same physical server, sharing CPU, memory, and disk. Cheap, simple, limited.
- VPS (Virtual Private Server): a partitioned slice of a server with dedicated resources, your own operating system instance, root access. Mid-tier in price and capability.
- Dedicated server: a whole physical machine. High cost, maximum control, rarely the right answer outside of specific workloads.
- Managed cloud: services like AWS, DigitalOcean App Platform, Cloudways, Vercel, or Platform.sh that abstract the server entirely. Pay for what you use, scale automatically, no operating system to manage.
The trap is choosing on price alone. Shared hosting is often £5 a month and dedicated is often £200 a month, so the temptation is to start with shared and “upgrade later”. For some sites this is correct; for others it is the most expensive decision available because the constraints of shared hosting cost the business in ways that do not show up on the hosting bill.
When Shared Hosting Is the Right Choice
Shared hosting is genuinely fine — even optimal — for a defined set of cases. Static or near-static brochure sites with low traffic. Personal projects. Small WordPress sites for a small business that does not depend on the site for revenue. Hobby experiments.
The honest indicators that shared hosting is appropriate: traffic under a few thousand visits a day, no e-commerce or only very low-volume e-commerce, no user-generated content at scale, no compliance requirements, and an honest acceptance that the site can go down for an hour without significant business impact.
The shared hosting market is competitive and the products are decent at the price point. SiteGround, Krystal, and 20i in the UK offer shared plans that handle the workload they are designed for. The mistake is using them for workloads they are not.
When Shared Hosting Becomes the Wrong Choice
Shared hosting starts failing when one or more of these is true:
- Traffic spikes. Shared hosting shares CPU and memory across hundreds of sites. Your neighbour’s viral moment becomes your downtime.
- E-commerce volume. A WooCommerce store with thousands of orders a month starts hitting shared hosting limits — slow checkouts, occasional database timeouts, gradual degradation.
- Custom code or processes. If your app needs background workers, scheduled tasks at scale, or specific PHP versions and extensions, shared hosting is restrictive.
- Compliance requirements. PCI-DSS, GDPR-sensitive workloads, anything where you need to control what software is installed and who has access — shared hosting cannot guarantee any of this because it is shared by definition.
- Performance is a competitive issue. If page speed matters for SEO or conversion, shared hosting is rarely fast enough.
The signal in practice is that the team is spending time worrying about the site rather than thinking about the business. When you find yourself checking whether the site is up before sending out an email campaign, shared hosting has stopped being the right answer.
The VPS Sweet Spot
For most growing business sites and small-to-medium custom applications, a VPS is the right answer. You get dedicated resources, root access, the ability to install whatever software you need, and a price that is moderate — typically £20–£100 a month depending on size.
The trade-off is responsibility. A VPS is unmanaged by default. You — or your developer — are responsible for keeping the operating system patched, configuring the web server, managing backups, and responding when something goes wrong. The hosting provider gives you a server and a network connection; everything else is yours.
This is fine if you have a development partner on retainer or in-house. It is a problem if “we have a VPS” really means “we have a server nobody is responsible for”. The mid-tier server with no ongoing management is the most common bad outcome — it works for two years and then drifts into security vulnerabilities and outdated software because nobody owned it.
A managed VPS (Cloudways, for example) splits the difference: dedicated resources, root access if you want it, but with someone else handling patching, backups, and monitoring. The price is higher than raw VPS but lower than dedicated, and the headache reduction is significant for most businesses.
When Dedicated Servers Make Sense
A whole physical server is the right answer in a narrow set of cases: very high traffic that consumes the resources of a substantial machine, workloads that need specific hardware (lots of disk, specific CPU types, GPUs), compliance requirements that demand physical isolation, or specific licence requirements that price per CPU.
For most businesses, the situations that look like they need a dedicated server are usually better solved with a properly sized VPS or with cloud hosting. The marketing case for dedicated — “the whole server is yours” — is real, but the practical case is rare. If you cannot describe specifically why your workload needs a dedicated machine, you probably do not need one.
When Managed Cloud Wins
Managed cloud is the right answer when you want to think about your application, not about servers. Vercel for Next.js sites. AWS, Render, or DigitalOcean App Platform for Laravel and Node. Platform.sh for almost anything.
The advantages are real: automatic scaling under load, no operating system to patch, deploys that are predictable, integrated monitoring, and infrastructure-as-code for repeatability. The disadvantages are also real: the pricing model can be hard to predict (especially with AWS), vendor lock-in is meaningful, and the abstractions sometimes leak — when something goes wrong you can find yourself debugging through a layer of platform you do not control.
A concrete example. A client building a Laravel SaaS chose DigitalOcean App Platform for the launch — auto-scaling, no server admin, integrated database. The cost at launch was around £40 a month. As traffic grew, the cost crept upward, predictably matching usage. At a certain scale (around five times the launch traffic), it would have been cheaper to move to a managed Kubernetes setup or a fleet of VPSes. The right time to make that move is when the cost crossover materialises, not earlier — early optimisation costs more in engineering time than it saves on hosting.
The Hidden Costs to Include
The hosting bill is rarely the dominant cost. The line items that get missed:
- Migration. Moving from one hosting model to another typically takes one to five days of work, depending on complexity. Plan for it.
- Operational overhead. Unmanaged VPS or dedicated requires someone to look after it. Budget the developer time honestly.
- Backups and restore testing. Backups are usually included. Testing that they restore is your job. Budget time annually.
- SSL certificates, domains, email hosting. Often forgotten until they expire and break production.
- CDN and adjacent services. Cloudflare, image CDNs, edge caching. These are often more important to performance than the underlying hosting.
- Downtime cost. If the site being down for two hours costs the business £5,000 in lost revenue, the hosting tier needs to reflect that — cheap hosting on a revenue-critical site is a bad bet.
Total cost of ownership for hosting is usually three to five times the visible hosting bill once these are included. That changes the comparison between options significantly.
Common Mistakes
- Picking on price alone. £5 a month versus £50 a month looks like an obvious choice until the £5 site goes down on launch day. The hosting tier must match the workload.
- Outgrowing shared hosting in slow motion. Shared works fine until it doesn’t, and the symptoms creep in — slower pages, occasional errors. Plan the upgrade before the site is already breaking.
- Buying a VPS without owning it. Unmanaged VPS with no responsible party is the worst hosting outcome — paying for capability that is not being managed.
- Treating cloud hosting as automatically cheaper. It can be, at low volumes. At high volumes, cloud bills compound. Budget for the volume you expect, not the volume you have today.
- Forgetting backups exist. “The host has backups” is not the same as “we have tested that the backups restore.” They are not equivalent.
- Migrating reactively rather than strategically. The worst time to choose new hosting is during an outage. Make the decision when the business is stable and the team has time to plan it properly.
What Good Looks Like
A well-chosen hosting setup matches the workload, has a clear owner responsible for the server (whether that is internal, a managed service, or a retainer partner), and is sized for the traffic you actually have with headroom for growth. The total cost — hosting plus operations plus backups plus CDN — is understood. The site or app runs at a speed the team is comfortable with, downtime is rare, and when something does go wrong, someone knows exactly what to do. Six months in, the team is not having to think about hosting at all because it is doing its job invisibly.
Next Steps
If you are about to commission a custom application and the hosting question is wrapped up with the build, How to Scope an Internal Software Project covers shaping the project. If hosting is the entry point to a broader infrastructure decision, Hosting and Infrastructure explains how we approach this for clients. For tailored advice on the right hosting tier for your application, get in touch.