You’ve done the work. Your WordPress site is clean, the plugins are minimal, and you’ve spent hours trying to WordPress optimize speed on the front end. Yet, when you load your site, it still feels like it’s wading through treacle.
It is a frustrating place to be, but you are not alone. Once your build is clean, the hosting environment becomes the loudest remaining variable. It is also the part of the equation that most business owners understand least.
This post is a practical buying checklist, not another guide on tweaking settings. We are going to look at the server-side factors that move the needle more than anything you can install after launch. We will walk through why a clean build can still feel slow, explore what time to first byte (TTFB) actually tells you, and explain the mystery of PHP workers. We will also cover server-level caching, why the physical distance to UK customers matters, and how to test a host before you commit. Finally, we will look at when paying more for better hosting is actually cheaper than paying for another rebuild.
Why a clean build can still feel slow
You’ve spent a fortnight on a hand-coded theme, trimmed the plugins, and the site still feels like it’s wading through treacle. That’s the moment to look away from the page. The browser spends its first moments waiting for your server to answer, not reading your markup, so a slow handover sinks everything after it. Lighthouse gives server response time its own audit, “Reduce server response times”, which tells you the fix lives at hosting level, not in your settings. A rebuild makes the front end lighter; it doesn’t make the server reply faster. If page-builder bloat was part of your old problem, our guide to why page builders slow your WordPress website down and the related category more on page builder speed explain it. But most buyers compare hosts on price, disk space and unlimited badges, while the numbers that decide real speed never appear on the sales page.
What WordPress speed optimisation can and cannot fix
If the build is already lean, it’s tempting to reach for another plugin. Caching plugins, image compression and script deferrals do genuinely help, because they reduce how much work the browser does after the response arrives. None of them make the server answer sooner, though. A slow time to first byte stays slow no matter how many plugins you stack on top.
If you’ve already run a speed plugin and the scores barely moved, treat that as evidence: the bottleneck sits upstream of WordPress. Google Search Central maintains its Core Web Vitals guidance, so speed is worth taking seriously, but no plugin removes server delay. Before you migrate, record a baseline: what your current host does well and badly, plus the numbers to prove it. Otherwise you’ll have nothing to compare the new host against. If you’re unsure whether plugin bloat or server delay is the culprit, our guide to speeding up WordPress without plugins helps you separate the two.
Time to first byte: the number that exposes a slow host
That delay has a name, and it’s the first place to look.
Meet TTFB
Time to first byte measures the gap between the browser sending its request and receiving the first byte back. Everything else on the page, every image, every script, happens after that number. If the server is slow to answer, the browser is just waiting.
So what’s a good score? Rather than trust a figure quoted in an old blog post, check the current guidance in web.dev’s dedicated TTFB article.
Test a real page
Test a page with real content, not the homepage. Note whether the value changes between the first visit and a repeat visit. A value that stays high on repeat visits points at server compute or database queries. One that only spikes on the first load is more likely to be a caching question.
Then ask the host plainly: what TTFB do you expect on a standard WordPress plan, and can I see a test site before I pay anything?
Next, let’s look at the number that never appears on the pricing page.
PHP workers: the queue nobody mentions on the pricing page
TTFB tells you how long a single request waits. PHP workers explain why that wait can grow. WordPress runs on PHP, and your host allows a fixed number of PHP processes to run at once, often set by a PHP-FPM limit. When they are all busy, new requests sit in a queue behind them.
Shared plans tend to cap this tightly. That is why a site can feel instant with one visitor and sluggish with a dozen. Ask how many PHP workers you get, whether that figure applies per site or per account, and whether it can be raised without changing plan.
Shops and membership sites hold processes open for longer, so the same worker count behaves differently on WooCommerce than on a brochure site. Be wary of hosts who quote only CPU and memory. Worker count is what shows up as “the site went slow this afternoon.” If your build is already lean, a page-builder audit is a separate job; start with why page builders slow your WordPress website down. Next, check what gets cached before WordPress wakes up.
Server level caching: what should be handled before WordPress wakes up
How close are the servers to your UK customers?
Caching sorted, the next question is geography. A server in the United States adds round trip time to every request a UK visitor makes before your content is even considered. Treat that as reasoning rather than a measured guarantee: no single number fits every site, and a CDN partly offsets distance for static files. So ask where the hardware physically sits, which UK or European region your plan uses, and whether you get a choice. Check the host’s own network or status pages, and be suspicious of a UK hosting badge that only means a UK billing address. Then test it: run the same page through a UK-based speed test and one from further afield, and compare. If the gap is wide, keep looking. Next, we will look at how to test a candidate host properly before you commit.
How to test a host before you commit
Location matters, but you can’t feel it from a sales page. So test the response times yourself. Start by asking for a trial, a staging environment or a demo domain. Use it before you move a live site.
Bring a copy of your actual site and plugins, not a blank WordPress install. An empty site will flatter any host. Then measure the same page, the same way, on your old host and the candidate, ideally on the same day and at a similar time. Keep a written record, so the decision rests on numbers rather than how quickly the sales team replied.
Ask what you get on a trial plan, from access to support, and confirm you can check what you get with ongoing maintenance before you commit to anything. Next, let’s look at the first test worth running.
Step one: a website speed test on a page that actually matters
Start with the page that pays the bills. Open Google PageSpeed Insights and enter the URL of a service or product page that brings in enquiries, not the homepage. Homepages flatter a site.
You’ll get two sets of results. Field data comes from real Chrome visitors over the past 28 days, so it reflects actual experience. Lab data is one simulated run on Google’s hardware. They can disagree, and that’s normal.
Never pitch a lab score from an empty test site against field data from a live site with traffic. Different things entirely. Ask the host to run the identical test on their demo site: same page, same tool, same conditions.
Finally, don’t chase the single score. Scroll to server response time. If that line is red, everything below it is damage limitation, and no amount of squashed images will rescue it.
Step two: measure time to first byte more than once
PageSpeed Insights gives you a single snapshot. TTFB, though, changes with load, so measure it on at least three separate occasions at different times of day before you draw any conclusion. Where a tool lets you pick a test location, choose the UK, since that reflects the customers you are trying to reach.
Then compare a cached load with an uncached one. If the cached version is fast and the uncached one crawls, a cache is covering for a server doing genuine work slowly. Note the worst result as well as the typical one, because visitors remember the slow outliers, not your average.
Finally, repeat the exercise a few weeks after the migration, once real traffic has arrived. Day one on an empty site flatters any host. Same page, same tool, same UK location.
When paying more for hosting beats another rebuild
Once you have those TTFB numbers, put the annual cost of the better hosting plan beside the one-off cost of another rebuild. Include your own hours as well as developer fees.
Then ask what a rebuild would actually change. If the build is already clean, a new theme will not fix a server that takes too long to respond.
Add the cost of the problem continuing: lost enquiries, slower admin, and optimisation rounds that never quite land.
A hosting uplift is recurring. A rebuild is a one-off. Both answer the same problem, so judge them that way, not on which feels cheaper today.
Finally, set a review point three months after migrating and use your own measurements to decide whether the extra money earned its keep. For a plain-English second opinion on where your slowness sits, see the questions people ask.
What to do next
Before anything changes, save your current TTFB and PageSpeed Insights numbers. Note both field and lab results, plus the pages you tested, so you have a fair baseline later.
Send four plain questions to two or three hosts: what TTFB do you expect on a standard WordPress plan, how many PHP workers do I get, which caching layers do you run, and where are the servers? Ask for a trial or staging site, then test it with your own content and plugins before you book a migration date. Questions people ask about WooCommerce performance are worth having ready if you run a shop.
If the build is already lean and the server is the bottleneck, moving host is usually the shorter route. Total Webz builds hand coded WordPress and WooCommerce sites without page builder or plugin bloat, and a free site check gives you a second opinion.
A month after moving, re-test the same pages with the same tools. Then judge whether the new plan earned its keep.
Conclusion
The takeaway is simple: a slow WordPress site isn’t always a build problem. Time to first byte, PHP worker limits, server-level caching, and server location are the foundations no amount of optimisation can fix. Test honestly before you commit, measure twice, and judge hosting on the numbers you record yourself. If the build is already lean, moving host is often the shorter, surer route.
Your next step: request a free site check from Total Webz, who build hand coded WordPress and WooCommerce sites without page builder or plugin bloat. Save your baseline today, ask hosts the four plain questions, and let real measurements, not guesswork, guide the move.