Guess your site’s speed score before you run a real test.
Answer 8 questions about how your site is actually built. You’ll get a heuristic score in the time it takes to read this sentence — zero URL fetching, zero waiting on a lab test queue.
This is a rule-of-thumb estimator, not a Lighthouse or CrUX report. It won’t catch what’s actually happening on your server. Treat the score as a “where do I stand roughly” gut-check, then confirm with PageSpeed Insights or WebPageTest before you act on it.
The 8 factors 0 / 8 set
Estimated score
Top fixes, ranked
How the heuristic weighs each factor
No single input decides the score. Each factor is weighted by how consistently it correlates with real-world load time and Core Web Vitals in published field data — not by guesswork ranking.
Straight answers
Is this the same as a Google PageSpeed Insights score?
No, and it isn’t trying to be. PageSpeed Insights actually loads your page in a real or simulated browser and measures Core Web Vitals. This tool asks you what’s true about your setup and applies fixed weights. It’s faster and needs no URL, but it can’t see mistakes you don’t know you’re making.
Why not just crawl the URL automatically?
Because a heuristic estimator that fakes a live crawl is worse than one that’s upfront about being a questionnaire. Real crawling needs a headless browser, waterfall analysis, and repeated runs to average out noise — that’s what the lab tools already do well.
My score seems too generous. What am I missing?
Likely three things this tool can’t see: third-party scripts (ads, chat widgets, analytics) stacking up after load, layout shift from web fonts, and mobile-specific throttling. Field data from PageSpeed Insights accounts for all three; this estimator doesn’t.
What’s a “good” score here?
80+ means your fundamentals are in decent shape. 60–79 means there’s one or two structural issues worth fixing this month. Below 60, the fix list on the right is doing real work — start from the top.
Does page weight matter more than request count?
In this model, yes — weight is weighted higher because a handful of huge images stalls a connection worse than many small cached requests over HTTP/2. That said, both matter, and fixing one rarely fixes the other.