Introduction
When measuring how fast your site feels to users, averages can be misleading. That’s where percentiles, and especially the 75th percentile (P75), come in.
What is a percentile?
A percentile is a way to understand how a specific value compares to all the other values in a dataset. It answers the question: "What percentage of users had a better (or faster) experience than this point?"
Let’s break it down with two common examples:
P50 is the 50th percentile. It is also known as the median. This means half of your users had a better (faster) result, and half had a worse (slower) one. It shows the "middle" experience.
P75 is the 75th percentile. It means that 75% of users had a better experience, and 25% had a worse one. This helps you understand what the experience looks like for users in the bottom quarter — those who are struggling.
So, P75 is not an average. It highlights the upper bound of what most users go through, and it's especially useful for spotting problems that affect a significant portion of your audience, even if they don’t pull the average down dramatically.
Think of P75 as the "worst experience among the best 75%", a critical threshold for performance.
Why P75
When it comes to measuring performance, not all metrics tell the full story. P75, the 75th percentile, has become the go-to standard, especially for web performance metrics like Largest Contentful Paint, Cumulative Layout Shift, Interaction to Next Paint and Perceived Load Speed. And here is why P75:
✅ More reliable than averages
Averages can be misleading. If most users have a fast experience but a few have extremely slow ones, the average will be dragged down and won’t reflect what the majority actually experienced. P75 ignores those rare outliers and focuses on the experience of 75% of your users. It tells you, “This is how bad it gets before a quarter of your users start having real trouble.”
⚠️ Highlights where users struggle
If your P75 is poor, it’s not just a few unlucky users — a large chunk of your audience is having issues. That’s a red flag. P75 helps surface performance bottlenecks that might be hiding behind a good average or median.
📊 It’s the industry standard
Google uses P75 in the Chrome UX Report (CrUX) and Core Web Vitals scoring. That means when your site is evaluated for things like SEO performance or Page Experience, P75 is the number that matters.
From Google Chrome team
“P75 CLS means 75% of users had that value or better—it doesn’t mean they all got that exact number.”
— Barry Pollard, Chrome Team
A Simple Example
Suppose you have 12 pageviews from real users on your website, and you want to analyze their Cumulative Layout Shift (CLS). First, sort the CLS values from lowest to highest. Next, determine the 75th percentile (P75) by identifying the pageview that represents 75% of the 12 pageviews, which is the 8th pageview in the sorted list below. This CLS value is your P75, as seen below.
In a similar way you can calculate any P on any real user metric.
How to use percentiles in Uxify
Uxify Experience uses P75 for real user monitoring by default. To change this, click the Filters button in the top-right and adjust the settings.
Conclusion
P75 shows what most users really experience, especially the ones who struggle.
If your P75 is in the red, your job’s not done