Small fluctuations in scores should never trigger panic reactions. Individual test results are almost meaningless due to strong variations. Only long-term trends, confirmed by regular checks, hold any real significance.

At each test time, not just a single test was conducted, but an average of five tests was taken. Many sources recommend this averaging approach for more reliable insights.

My test results show: Despite averaging multiple tests, the scores still vary significantly from day to day.

while other features, contrary to expectations, had no significant negative impact. This can drive you crazy if performance optimization is a declared goal.

To dig deeper, I tracked PageSpeed test results on multiple websites without releasing speed affecting new features over several weeks.

Why You Should Never Give Too Much Importance to Individual #PageSpeed & #Lighthouse Scores – A Thread.

In the past, we meticulously analyzed PageSpeed test results after releasing new features. Too often, performance optimizations didn’t improve the score as expected ...

#WebDEV #Frontend

I now use subgrid for every type of teaser with at least two centered texts that can vary in height. However, this and other examples have not been released yet.
I have never come across subgrid on others websites though 😅