Website testing tools compared
What to actually measure a site by, how lab data differs from field data, and why one measurement means nothing.
Two kinds of data that are easy to mix up
This distinction matters more than which tool you pick. Lab data comes from a simulation: the tool opens the page on a virtual device, on an artificially throttled connection, and measures. It's repeatable and good for before-and-after comparisons.
Field data comes from real people who actually visited — on their devices, on their connections. It's the truth about what's happening, but it arrives with a delay and is averaged over twenty-eight days.
A gap between the two is information, not an error. It usually means your visitors have slower devices than the simulation assumes, or use mobile networks more than the test does.
Four tools that are enough
-
Google Search Console
The most important one, and free. Field data on Core Web Vitals, indexing, the queries your pages show up for. The only source that speaks to your real visitors.
-
PageSpeed Insights
Free, in the browser. Shows lab and field data side by side, with a concrete list of what to fix. Good for a first look and for talking to a contractor.
-
Lighthouse
The same engine, built into Chrome (the Lighthouse tab in DevTools). Runs locally, no limits, even on pages that aren't public yet.
-
WebPageTest
For going deeper: choice of location, browser and connection speed, a frame-by-frame film of loading, a request waterfall. Free at the basic tier.
How to test so the result means something
- In a mobile profile. Most visitors are on a phone, and that's where the differences are biggest.
- Three times, median. Not the first result, not the best one.
- On a page that actually gets traffic — not just the homepage. People often land straight on a product page.
- With cache cleared if you're measuring a first visit, without clearing it if you're measuring a returning one.
- Record the result with a date. Without a baseline you can't tell if anything improved.
- Don't compare your result to a competitor's without checking what's on their page. A page with no photos will always beat a shop.
What's not worth tracking
A single overall score. It's convenient to show a manager, and misleading: it can be raised without changing anything a user would notice, and a page that's subjectively slow can still score well.
"Visibility" as measured by the tool vendor's own tool. Every one calculates it differently and none of them is a source of truth — unlike Search Console, which shows actual impressions and clicks.
Request counts in isolation from their weight and order. Forty small files from the same server can be faster than five huge ones.
Questions
- Where do I start if I have none of these tools?
- With Search Console. Verification takes a few minutes, it's free, and it's the only place that shows what real visitors to your site actually experience.
- Do I need a paid tool?
- Not for measuring speed. Paid tools make sense for tracking rankings across many keywords and analysing competitors — that is, for ongoing work, not a one-off fix.
- The score is good but the site feels slow. How?
- Usually it's response time to interaction (INP), which a lab test measures poorly, because nobody clicks anything in it. Look for that data in Search Console.
Want us to measure it for you?
Send the address. A measurement and a short priority list are free.
Write to us