People notice their hosting exactly twice. The first time is at sign-up, because the site has to live somewhere. The second is when the site goes down mid-sale, the checkout form refuses to open, and the provider’s support replies “try again in an hour.” Between those two moments hosting is invisible — which is precisely why it usually gets picked on a single criterion, price. That is a decision that costs far more later than the few dollars a month you saved.
At Netloria we build websites end to end, and almost every time a client arrives with a site already running, we see the same story: hosting chosen at random, off the first banner in search. Here is how to work it out yourself — how hosting types actually differ, which specs to read before you pay, and which “advantages” in a plan description mean nothing at all.
Hosting is renting space on a server that stores your website’s files and database and hands them to a visitor’s browser every time someone opens a page. How fast, stable, and close to your audience that server is determines how quickly the page opens — and whether it opens at all during a traffic spike. A domain is your address; hosting is the building itself. They are bought separately and judged by different criteria, so it pays not to confuse them.
Now about the “unlimited” disk space and “unmetered” traffic splashed across nearly every provider’s landing page. Unlimited does not exist in the physical world: behind the pretty word sits a “fair use” clause in the terms of service. The moment your site consumes more than a cheap plan budgets for, it is either quietly throttled or you are politely asked to upgrade. So “unlimited” is not a spec but a promise, and the part worth reading is the fine print underneath the headline, not the headline.
The type of hosting decides how much of the server’s resources are yours and how dependent your site is on its neighbours. This is the main fork in the road; everything else is secondary. Here is how the core options compare, with rough entry prices for orientation.
| Type | What it really is | Best for | Rough entry price |
|---|---|---|---|
| Shared | Your site shares one server with dozens of others. Resources are pooled, control is minimal. | Landing page, business-card site, small blog with modest traffic. | from ~$3–5/mo |
| Managed WordPress | Same idea, but the server is pre-tuned for WordPress: caching, updates, security handled for you. | A corporate site or blog on WordPress with no in-house admin. | from ~$5–10/mo |
| VPS | A dedicated slice of a server with guaranteed resources and full control over settings. | Online store, high-traffic site, several projects on one server. | from ~$10/mo |
| Dedicated | A whole physical server for you alone. Maximum power and maximum admin responsibility. | Large store, high-load service, strict security requirements. | from ~$30/mo |
The logic is simple. The higher the tier, the more isolated resources and control you get — but the more technical responsibility lands on you. A small business with one business-card site gains nothing from a $30 dedicated server; that is paying for power with nothing to run on it. An online store that takes ten times its usual traffic on Black Friday, however, will be knocked offline by a three-dollar shared plan at the exact moment it should be earning. Match the type to your load and growth plans, not to the “cheapest available” line in the budget.
Cloud hosting deserves its own note. Its main advantage is that resources scale with load automatically, and you pay for what you actually use. That saves projects with unpredictable peaks: seasonal sales, a post that goes viral, an ad campaign that suddenly lands. The downside is that the bill is unpredictable too, so cloud makes sense once you already understand your traffic profile.
Once the type is settled, the comparison of concrete parameters begins. Here is what genuinely affects how your site runs — and what most people check last, wrongly.
1. Uptime — the share of time the server is online. This is the single most important number, and it is shyly buried in the fine print. The accepted norm sits between 99.9% and 99.99%. The gap looks cosmetic, so translate it into hours of downtime per year: 99.9% is almost nine hours offline, while plain 99% is over 87 hours — three and a half days a year when your site simply does not work. A provider that never states a guaranteed uptime is a red flag.
2. Server response time (TTFB). This is the delay between the browser’s request and the first byte the server sends back. An acceptable TTFB is up to 600ms, and an ideal one is under 200ms. You cannot cache your way past a slow server — that is a ceiling, not a setting.
3. Server location. Physical distance between server and visitor is real milliseconds of delay. Intercontinental data hops can add well over a hundred milliseconds before the server even starts processing anything. Keep the base server near your primary market; if your audience is global, a CDN serves content from the nearest edge.
4. Disk type. NVMe beats ordinary SSD, and SSD is many times faster than legacy HDD. You hear it immediately in database speed, especially on a store with thousands of products. If the plan does not name the disk type, it is almost certainly the cheapest option.
5. PHP and server software version. Modern WordPress wants a current PHP release; a host stuck on five-year-old versions is both slower and less secure. A good provider lets you switch the PHP version yourself in the panel rather than through a support ticket.
6. Backups. The question is not “do they back up” but “how often, and can I restore it myself in a couple of clicks.” A daily automatic backup with self-service restore is the thing that saves your business one day. We walked through that exact scenario in our guide on what to do when your site gets hacked — a fresh backup is what rescues the situation most often.
7. Support. A cheap plan often means support “sometime within a day, by email.” When the site is down, a day is an eternity. Check for 24/7 chat and the language they answer in before you pay, not during an outage.
This is where it gets interesting, because “slow site” stays abstract until you turn it into lost revenue. And the numbers exist, and they are harsh. Google’s research found that the probability of a bounce rises by 32% when load time goes from one second to three. A slower server does not “lag a bit” — it systematically pushes a third of people out before they have seen your content.
Pingdom data from the same body of research makes the trend even plainer: a page that loads in one second sees a 7% bounce rate, three seconds gives 11%, and five seconds already gives 38%. None of that is about design or copy — it is pure infrastructure, which is exactly what your hosting decides.
A third of your prospects vanish while a slow server gathers its thoughts. A measured 32% is not about design.
Now conversions. A Portent study recorded that an e-commerce conversion rate falls from 3.05% at a one-second load to 1.08% at five seconds. In plain terms, slow hosting can cut your sales by nearly two-thirds, and you will never see it as a line item — just fewer orders. That is why we at Netloria tell clients to treat speed not as a developer’s whim but as a funnel with a slow leak.
There is one more level where slow hosting hurts: search. Google has long treated load speed as a ranking signal, so a site on a weak server loses visitors at the door and also ranks lower, missing people before the click. That is a double penalty — fewer visitors reach you, and some who do give up mid-load. Hosting quietly moves both numbers at once, and no ad budget makes that back; you simply pay more per customer.
Hosting is the first and cheapest lever on that speed. Everything else — caching, image compression, database cleanup — matters too, but no optimization lifts a site above the ceiling the server sets. And to know whether speed is actually costing you clients, first put your measurement in order — we covered that in the piece on website metrics that actually matter.
The question we get most is “local or foreign host,” and there is no universal answer — it depends on where your customers are. If your audience sits in one region, a server in or near it gives the best response time and support in a language you understand. If you sell across the EU or US, keep the server close to those markets or take hosting with a CDN so content is served from the nearest point to each visitor.
There is a second, often-forgotten angle: billing and paperwork. As a web studio based in Ukraine, we help clients weigh a provider not only on specs but on how it invoices, what documents it issues, and which payment methods it supports. That is not a technical criterion, yet it is sometimes the deciding one.
Plenty of people cling to bad hosting for years simply because they fear the move — it feels complex, and surely the site will collapse mid-process. In reality, migrating a site to another server is a routine job that most decent providers do for free and by themselves when you switch to them. Technically it is copying files and the database to the new server, testing on a temporary address, and flipping the domain’s DNS records — and done in the right order, visitors never notice anything changed. So a rushed choice made once is not a life sentence.
The signal that it is time to price alternatives is simple: the site is regularly unreachable, support answers in days, and the only fix offered for a slow server is a pricier plan with no explanation. When that happens, moving pays off faster than you would expect. In our experience, migrating a mid-size corporate site takes a few hours of work, not the weeks everyone dreads.
One catch people genuinely get burned by: register the domain in your own name, not the host’s.
Before you enter card details, run through these. If half of them have no clear answer on the provider’s site, keep looking.
Hosting is the case where the right choice stays invisible for years and the wrong one reminds you of itself at the worst possible moment. A cheap plan with no guarantees is not a saving — it is a deferred risk.
If you are ordering a website and have no wish to dig into TTFB, PHP versions, and plan fine print — that is fine; a studio exists for exactly that. At Netloria we select and configure hosting for the specific project during development: so the server fits the load, sits close to your audience, and does not become the bottleneck after launch. See how it works with our corporate website development service, where hosting is already part of the solution rather than left to you “for later.”
