Site Speed Metrics – The Minimum You Should Consider

We all know site speed matters – for SEO, user experience, and conversions. But start comparing results from tools like PageSpeed Insights, GTmetrix, and Pingdom, and it’s easy to get lost in the numbers…

This week, I want to simplify it: here are the minimum site speed metrics you should actually pay attention to (and you should ALWAYS check before live) – and the tools I use to measure them reliably.

Why You Shouldn’t Rely Solely on PageSpeed Insights

PageSpeed Insights (PSI) is great for diagnosing specific performance issues, but it’s not a reliable benchmarking tool because it’s really a mix of two different data sets.

The first is Field Data, which is a 28-day rolling average of how real Chrome users experience your site. The second is Lab Data, which is just a one-off test run in a controlled environment (Google’s Lighthouse engine). This is why your score might jump around even when you haven’t changed anything.

That means your score isn’t a single number – it’s an average of thousands of visits, updated slowly, influenced by device type, geography, and connection quality. If your audience includes people on 4G in rural areas, your average might never look “fast.”

It’s brilliant for spotting real world issues (like uncompressed images, slow fonts, or render-blocking scripts), but it’s terrible for checking that your fixes are actually going to work. 

The Tools I Actually Use (and Why)

For consistent, repeatable results, I use Pingdom and GTmetrix side by side.

Pingdom keeps it simple – it measures total load time, page size, and number of requests. These are easy to understand and compare month to month.

GTmetrix adds depth. It gives me the key web vitals – things like Largest Contentful Paint (LCP), Total Blocking Time (TBT), and Cumulative Layout Shift (CLS) – without the inconsistencies of PSI’s real-world data.

Together, they tell me both how fast a site loads and how smoothly it behaves once it does.

The Minimum You Should Pay Attention To

You don’t need a tonne (yes, that’s how it’s spelt in the UK) of metrics to know whether a site’s healthy. Focus on these fundamentals:

• Load Time: aim for under 2.5 seconds
• Page Size: under 2MB is a solid target
• Requests: keep them below 100
• LCP: under 2.5s, ideally 2s
• TBT: below 300ms
• CLS: under 0.1

If you’re seeing these numbers consistently, then you’re in good shape…

Quick Agency Tip 

When you show speed data to clients, don’t confuse them with jargon (and definitely don’t show them the waterfall graph 😂). Pick three clear metrics – overall load time, LCP, and either TBT or CLS – and present them as proof that the site feels fast and stable to users.

And when you run tests, don’t just check the homepage. Test a selection of a product/service page, a blog post, and a template-heavy landing page too. They often behave very differently, and it’s crucial to know that wherever a user enters the site, they get a good experience. 

One more thing…

You don’t need a perfect 100/100 score. You need a site that loads quickly, doesn’t jump around, and feels effortless to use.By combining Pingdom and GTmetrix for consistent testing, and using PageSpeed Insights purely for diagnosis, you’ll have a clear, repeatable system for measuring real performance – not just chasing arbitrary numbers.