UX Rating is Calibre’s 0 to 100 score for how fast, stable and responsive your Site is for real visitors. A UX Rating of 80 means 80% of your visitors’ experience was good.
It uses four metrics:
- Largest Contentful Paint (LCP)
- Cumulative Layout Shift (CLS)
- Interaction to Next Paint (INP)
- Time to First Byte (TTFB)
Each result is rated good, needs improvement or poor against that metric’s thresholds. UX Rating is the share rated good.
Calibre calculates UX Rating for both Real User Monitoring and CrUX, in slightly different ways, explained below. Synthetic tests aren’t real visits, so they don’t have a UX Rating. Use the Lighthouse Performance Score instead. It’s scored differently, so don’t expect it to match UX Rating.
In Real User Monitoring#
Each page view reports up to four metrics, and each is rated good, needs improvement or poor. Calibre pools every rating in the selected date range and filters:
UX Rating = good ratings ÷ all ratings × 100For example, 1,000 page views. Every page view reports LCP, CLS and TTFB, but only 600 had an interaction, so only 600 report INP:
| Metric | Ratings | Good |
|---|---|---|
| LCP | 1,000 | 800 |
| CLS | 1,000 | 900 |
| INP | 600 | 420 |
| TTFB | 1,000 | 760 |
| Total | 3,600 | 2,880 |
2,880 ÷ 3,600 × 100 = 80. UX Rating is 80.
- It is not a pass rate per visit. A page view with one slow interaction adds one poor rating and three good ones. It is not counted as a failed visit.
- Missing metrics are left out. A page view with no interaction has no INP, so it adds three ratings, not four. Nothing is counted as poor because it wasn’t measured.
- Busier metrics carry more weight. Because ratings are pooled, a metric observed on more page views counts more. INP is usually observed less often than LCP, CLS and TTFB.
In CrUX#
CrUX doesn’t provide individual page views or a UX Rating. For each metric it gives the share of page loads that were good, needs improvement and poor over 28 days. Calibre combines the good shares of the metrics CrUX has data for, with each metric counting equally:
UX Rating = sum of good shares ÷ number of metricsFor example:
| Metric | Good share |
|---|---|
| LCP | 90% |
| CLS | 80% |
| INP | 70% |
| TTFB | 60% |
| Total | 300% |
300% ÷ 4 metrics = 75%. UX Rating is 75.
If CrUX has no data for a metric, it is left out. With no INP data, the same Site would be (90% + 80% + 60%) ÷ 3 = 76.7%, so UX Rating is 77.
Because RUM and CrUX weight metrics differently, and measure different visitors, the same Site can have a different UX Rating in each. See why your numbers differ.
Good, Needs Improvement and Poor#
Calibre grades each UX Rating as Good, Needs Improvement or Poor:
| UX Rating | Grade |
|---|---|
| 90 or above | Good |
| Above 50 and below 90 | Needs Improvement |
| 50 or below | Poor |
Related reading#
- UX Rating: where to find it and how to improve it
- Metric thresholds
- Why your Synthetic, CrUX and RUM numbers differ