How accurate is the data?
Data is based on real users, so it doesn't get any more accurate than that. There could be discrepancies though. For example in the following cases:
- your website or webshop's performance is very bad. As a result, not all users might be tracked. You might then see better data then it really is (as those with really poor UX could not be tracked);
- users could have JavaScript disabled;
- Not all metrics are supported by all browsers, which could skew gaps between FCP (good browser support) and LCP (not supported by all modern browsers) for example;
- Even when it comes to a single metric, there could be differences amongst browsers. Such as the FCP metric;
- When having a lot of (headless) bots, they are actually executing the page. Lighthouse is doing this as well, for example. When regularly doing synthetic tests, you could choose to exclude this data, or test with pages which aren't being tracked;
- When currently using performance hacks, data won't be as trustworthy. We encourage you to not use any hacks to get a better understanding of real UX. In the end, these metrics were introduced not to annoy us, but to get insights or real UX of our very own audiences.