Two owners asked me about website speed in the same month. One ran a manufacturing company whose site took eleven seconds to load on a phone and was spending $3,000 a month on Google ads pointing at it. The other ran a consultancy with a fast, clean site and a score of 84, and wanted to know how to get it to 95.
The first one had a real problem worth thousands of dollars. The second one was avoiding a harder conversation about what his website actually said. Both had read the same advice online.
What Google actually measures, in plain words
Google publishes three timing checks for web pages. The names are unhelpful, so here is what each one is asking.
How long until the main thing appears? This is the big headline or hero image at the top of the page. Google's target is under two and a half seconds. This is the one that matters most, because a visitor staring at a blank white screen has no idea whether your site is loading or broken.
How long until the page responds when I tap something? If a visitor taps a button and nothing happens for half a second, the page feels broken and they tap again. Google's target here is under 200 milliseconds, which is a fifth of a second.
Does the page jump around while it loads? You go to tap a link, an image finishes loading above it, everything shifts down, and you tap the wrong thing. This one is measured as a score rather than a time, and lower is better.
Google uses these as a ranking signal, and has been clear that it is a small one compared to whether your page actually answers the question. A fast page with thin content does not outrank a slow page with the right answer. Speed breaks the tie between two pages that are otherwise equally useful.
Where speed genuinely costs you money
There are two situations where I take load time seriously at a small B2B company.
Mobile visitors on ordinary connections. Most B2B site owners test their own website on a laptop in their office, connected to good wifi, with the page already cached in their browser from yesterday. That is the fastest possible version of your site and nobody outside your company experiences it. A site manager checking your services page from a job site on a phone with two bars is having a completely different day. Check your own analytics for the share of visitors on mobile. At most Vancouver B2B companies I look at, it sits between a third and half.
Paid traffic landing pages. This is where slow pages turn directly into wasted money. You pay for every click. If a page takes eight seconds and a third of those clicks leave before it renders, you paid for all of them and got two thirds of the traffic. Ad platforms also factor page experience into what they charge you, so a slow landing page costs more per click and delivers fewer visitors from each one. If you spend money on ads, the landing page is the first place to look, ahead of everything else on your site. Pair the speed check with a look at whether the page asks for the right thing.
Where speed is over-worried
Chasing a perfect score is where most of the wasted effort goes.
Moving from 85 to 95 on Google's testing tool rarely changes anything a visitor notices. Both pages feel fast. The gap between 30 and 70 is the one people feel. Past 85 you are spending real money and developer hours for a number that lives in a report.
Desktop scores also matter less than mobile ones, because desktop visitors are usually on better connections and bigger machines. If your mobile score is 40 and your desktop score is 92, fix the mobile experience and ignore the desktop number entirely.
Deep pages get over-worried too. Your blog archive, your privacy policy, your team bios: these pages get few visits and no enquiries. Spending a Saturday optimising them is a hobby. Focus on the homepage, the two or three service pages that get visits, and anything you are paying to send traffic to.
How to measure your own site in ten minutes
Two free tools, no account needed for the first.
Go to Google's PageSpeed Insights and paste in your homepage address. It gives you a mobile score and a desktop score out of 100, plus the three timing checks described above. Ignore the long list of technical suggestions underneath on your first pass. Look at two things: the mobile score, and the first timing number, how long until the main thing appears.
At the top of that report, look for a section showing data from real visitors over the past 28 days. That section is worth more than the score, because the score comes from one simulated test on a simulated slow phone. The real-visitor section tells you what people actually experienced. If your site has too few visitors, that section will be missing, which is itself useful to know.
Then run the same test on the page you send ad traffic to, and on your main service page. Three tests, ten minutes, and you now know whether you have a problem worth money.
The second tool is your own phone, on mobile data with wifi turned off, in a private browsing window so nothing is cached. Load your homepage and count. That is the experience you are actually selling.
The five fixes that produce most of the gain
In the order I would do them.
Image file sizes. This is almost always the biggest single problem and the easiest to fix. Someone uploaded a photo straight from a camera at 4 megabytes and the website is shrinking it on screen while still sending the whole file. Every image on a web page should be saved in a modern format called WebP, sized no wider than it appears on screen. A hero image should land under 200 kilobytes. I have seen homepages drop from nine seconds to three on this fix alone.
Too many tracking scripts. Over the years a site collects small pieces of code from advertising platforms, analytics tools, heat mapping, an abandoned chat provider, a form builder nobody uses. Each one is a separate download that delays the page. Open your site's tag manager or ask your developer for the list, and remove anything nobody has looked at in six months. Most small B2B sites can drop half of these with no loss.
Chat widgets. A live chat bubble is often the single heaviest item on a small business website, and on many of them nobody is staffing it. If you are not answering chats within minutes during business hours, remove it. If you are, set it to load after the rest of the page finishes rather than alongside it. Your developer will know the setting.
Font loading. Custom fonts download as separate files, and on a badly configured site the text stays invisible until they arrive. Limit yourself to two font weights, host them on your own domain rather than fetching them from elsewhere, and make sure text displays in a fallback font while the real one loads.
Hosting. Cheap shared hosting adds a delay before the page even starts. If your site's server takes more than about a second to respond before anything else happens, the host is the ceiling on everything else you do. Moving a small business site to better hosting is a half-day job and usually costs less per month than one Google ad click per day.
Those five cover most of what I find. A developer can do all of them in a day or two for a typical small business site.
When to stop optimising and fix the offer
Here is the test. Your page loads in under three seconds on a phone, your mobile score is above 75, and you are still getting one enquiry a month. Speed is no longer your problem.
At that point the questions change. Does a visitor understand what you do within five seconds of landing? Does the page name the kind of company you serve? Is there one obvious next step, or four competing buttons? Does your contact form ask for eight fields when three would do?
I see owners spend three months on performance work because it is measurable and comfortable. There is a score, it goes up, it feels like progress. Rewriting your homepage so a stranger understands your business is harder, less measurable, and worth far more. A conversion audit is the version of this work that actually moves enquiries, and turning visits into enquiries is where the money is, and the words on the page usually matter more than the milliseconds.
The takeaway
Run three tests: your homepage, your main service page, and any page you pay to send traffic to. Look at the mobile score and at how long the main content takes to appear. If you are under three seconds and above 75, you do not have a speed problem.
If you are above five seconds, start with image file sizes, then remove tracking scripts nobody reads, then deal with the chat widget. Those three fix most sites in a day.
Speed work has a ceiling, and past 85 you are decorating a report. The enquiries come from what the page says once it has loaded.