A WooCommerce store loads a server in a completely different way from a company site, and that is the main reason ordinary cheap hosting starts to choke it. WordPress hosting for WooCommerce comes down to three things in practice: at least 512 MB of PHP memory per process, several parallel PHP processes (workers) and a fast NVMe or SSD disk under the database. The rest - LiteSpeed, control panel, backups - matters, but comes second. If the store slows down with 20 people on it, or the cart “gets lost” mid-campaign, it is almost always workers and memory that are missing.
Below I break it into numbers and into the signals a store owner can see without touching server configuration.
Why a store needs more than a brochure site
A company site is mostly ready-made HTML pages generated once - the server sends them from cache and computes practically nothing. A store cannot work that way. The cart, checkout and customer account have to be computed fresh every time, because everybody sees something different in them. WooCommerce adds its own database tables, customer sessions, stock and shipping calculations, and calls to payment gateways.
The practical conclusion: the more people are simultaneously in the cart and at checkout, the more CPU usage grows - not bandwidth. That is why a plan advertised with “unlimited transfer” solves nothing for a store; the limit is compute power and the number of parallel PHP processes.
How much a WooCommerce store needs - concrete thresholds
Treat the values below as a working minimum, not an ideal. I check them in every store diagnosis.
| Parameter | Minimum for a small store | Store with traffic and campaigns |
|---|---|---|
PHP memory (memory_limit) | 512 MB | 768 MB - 1 GB |
| PHP version | 8.1 | 8.2 or 8.3 |
| PHP processes (workers) | 4-6 | 10 and up |
| Database | MySQL 8 / MariaDB 10.6+ | the same, on a separate NVMe disk |
| Disk | SSD | NVMe |
max_execution_time | 120 s | 300 s (product imports, invoices) |
| Backup | daily, off-server | daily + before every update |
Two rows in that table are underrated. PHP workers are the number of requests the server truly handles in parallel - with two workers, the third customer at checkout simply queues. max_execution_time decides whether a wholesaler price list import or invoice generation finishes or breaks off halfway.
SSD, NVMe and LiteSpeed - what actually changes
A fast disk and a fast web server are two different improvements that work in different places in a store.
- NVMe instead of ordinary SSD speeds up the database - product listings, filters and the orders panel. You feel it most in store administration, where nothing can be cached.
- WordPress hosting with LiteSpeed mainly makes sense because it pairs with the LiteSpeed Cache plugin and object caching - it can serve product pages without starting PHP. The alternative (Nginx with FastCGI Cache and Redis) gives a very similar result, so the word “LiteSpeed” on an offer is not an argument on its own.
- Object caching (Redis) is worth asking about explicitly. In a store with a large catalogue it usually gives a bigger jump than adding CPU, because it takes repeated queries off the database.
What must never be cached: the cart, checkout, customer account and the “my orders” page. If hosting caches everything with no exceptions, customers start seeing each other’s carts. That is one of the most embarrassing store failures, and it comes from one badly configured rule.
Shared, VPS or managed - and what it costs
Hosting price lists change every year and differ between providers, so I give orders of magnitude on the Polish market rather than exact figures. Check the current offer before deciding.
| Type | For whom | Cost order (annual, net) |
|---|---|---|
| Ordinary shared | a company site, not a store | a few hundred zloty |
| Shared “for WordPress”, NVMe, higher plan | a store up to a few hundred products, steady traffic | a few hundred to about a thousand zloty |
| Managed WordPress / WooCommerce | a store where you do not want to touch the server | from about a thousand zloty upwards |
| VPS with management | large catalogue, integrations, B2B, custom APIs | most expensive, plus the cost of administration |
The rule I repeat to clients: hosting is the cheapest element of a store and the worst place to save. The difference between the cheapest plan and a sensible one is usually the cost of one or two orders a month. What exactly you lose on the bottom shelf, I set out in the piece on cheap and free WordPress hosting.
A VPS with no management is a separate matter. The server itself can be cheap, but somebody has to update the system, watch certificates and react when PHP stops responding at 2 a.m. If you do not have that person, managed comes out cheaper than a VPS plus an outage on Black Friday.
Signs your hosting is no longer enough
You do not need to measure anything complicated. These symptoms are enough:
- The WordPress dashboard is slower than the storefront - the classic lack of power, because admin cannot be cached.
- The site slows down exactly when there is traffic - in the evening, after a newsletter, during a campaign.
- 502 or 504 errors when adding to cart - the request queue is overflowing.
- Product imports or stock updates break off halfway - script execution time too short.
- Order emails arrive late or the host blocks sending.
- The host sends warnings about exceeding process or load limits - that is no longer a warning, it is a deadline.
Before you pay for a bigger plan, though, it is worth ruling out the other side of the coin. Very often a store is slow not because of the server but because of too many plugins, a heavy theme and uncompressed images - I break that down in the piece on why a WooCommerce store runs slowly. A stronger server under an untidied store buys a few months of improvement, and then the problem comes back.
Multisite, “unlimited” and other traps in offers
WordPress multisite hosting lets you run many sites on one installation. For a network of blogs it can be handy; for stores, rarely. WooCommerce on multisite complicates payments, integrations and updates, and an outage takes all the stores down at once. If you run two stores, two separate installations almost always work out better.
What else to view critically:
- “Unlimited” everything - there are always limits, just hidden in the terms as process or CPU load limits.
- A backup “included” with no stated frequency or location - a copy on the same server as the store is not a backup. How to set it up sensibly, I described under WordPress backups.
- No staging environment - in a store you do not test updates on production.
- Email-only support - when a store is down, response time matters, not a pretty form.
- “Free” migration done by an automated tool - it works until the store has orders in progress.
How to move to better hosting without losses
Moving a store is not dragging files across. The order I use:
- A full copy of files and database, verified by restoring it, not just downloading it.
- Standing the store copy up on the new server, without changing the domain, and testing: test purchase, payment, email, invoice.
- Shortening the DNS record’s time to live a day before the move.
- Closing out orders in progress and a short maintenance window, ideally in your weakest sales hour.
- Switching the domain, issuing the SSL certificate, and repeating the purchase and payment test on the target address.
- Keeping the old server alive for a few days as insurance.
Point two is the key one. The most common loss during a migration is not files but orders placed “in between” servers - those that landed on the old address while some customers already had the new one.
When to outsource this
If the store is slowing down and you do not know whether the server or the store’s internals are to blame, I start with a free diagnosis: I look at real response times, load, PHP configuration and caching, then say plainly whether a plan change is enough or whether we have to go into store optimization. For bigger work - a migration, a caching rebuild, setting up a staging environment - you get a written quote before the start, with no blind work. Projects are usually billed fixed-bid in the 5,000-10,000 PLN net range, and day-to-day watching of the server, updates and backups falls under ongoing technical support - Basic from 1,000, Pro from 2,000, Premium from 3,500 PLN net per month.
Frequently asked questions
What hosting should I choose for a small WooCommerce store? A higher shared hosting plan dedicated to WordPress is enough, provided it has at least 512 MB of PHP memory, PHP 8.1 or newer, an NVMe or SSD disk and at least several parallel PHP processes. The cheapest “starter” plans are not suitable for a store, because their process limits are too rigid.
Is LiteSpeed hosting better for WooCommerce? It can be faster, but not by itself - the advantage comes from LiteSpeed working well with its own cache plugin and being able to serve pages without starting PHP. Nginx with caching and Redis gives equally good results. More important than the server name is how much memory and how many workers you have, and whether cart and checkout are excluded from caching.
What does WooCommerce hosting cost? A sensible store plan is usually in the range of a few hundred to over a thousand zloty net a year, with managed hosting and a managed VPS costing correspondingly more. Provider price lists change every year, so check the current offer before deciding - and above all check the PHP process limits, because they decide how the store behaves under traffic.
Is moving a store to different hosting risky? The risk is real but controllable. The biggest danger is orders placed while the domain is switching, and payments that stop working on the new address. With a full backup, a purchase test before the switch and a short maintenance window in your weakest sales hour, a store moves to a new server with no data loss and no downtime visible to customers.
Running a store where something does not work the way it should? I build and rebuild WooCommerce stores - from a single fix to reworking the whole sales flow. Tell me what you are dealing with and I will send back a scope and a price.