The author’s views are entirely his or her own (excluding the unlikely event of hypnosis) and may not always reflect the views of Moz.
In part one of this series, we talked about how Google and the web in general were not really ready for the Page Experience Update — Google’s CrUX data covered too few websites, the vast majority of which were not hitting the required thresholds. That was why, I suggested, the update had been so delayed and watered down.
In part two, we talked about the metrics themselves — their flimsiness, their arbitrariness, their openness to manipulation. This, too, I suggested, might be holding Google back.
However, the proof is in the pudding. Are Core Web Vitals, taken individually or as a whole, correlated with rankings? If so, is that any more true than it was before the Page Experience Update? In this third and final post of this series, we’ll see what the data tells us about the relationship between Core Web Vitals metrics and organic ranking performance.
Viewer discretion advised
This is, at most, a correlation study. There are many mechanisms by which something can be correlated with rankings without having directly influenced rankings.
For example, perhaps websites that take SEO seriously rank well, and also tend to work on their loading performance. If so, loading performance and ranking would be correlated even without any direct causal link.
We’ll talk through potential implications as we go, but please, proceed with caution!
Performance of passing vs. failing URLs
To start with, I decided to look only at the URLs that had CrUX data in the first place. You may remember from part two that, at the time of the update rolling out in August this year, that was some 38.3% of URLs. This is taken from the top 20 results for 10,000 MozCast keywords, across mobile and desktop device types.
Note that these URLs are all taken from the top 20, so it’s interesting that the averages are both well above the rank of 10.5 we’d expect. This is likely because higher traffic URLs are disproportionately likely to rank well, and also disproportionately likely to have CrUX data.