PageSpeed Insights Moves to Lighthouse 13.6: Scores May Rise Without Site Changes

PageSpeed Insights now uses Lighthouse 13.6.0, released on October 27. As Search Engine Journal reports, Lighthouse 13.6.0 changes how page loading is simulated. According to Chrome representative Barry Pollard, the estimated First Contentful Paint (FCP) and Largest Contentful Paint (LCP) metrics should be more accurate and generally produce higher scores. That means a page’s score may change even if nothing on the site has changed.

How Lighthouse recalculates page loading

Lighthouse loads a page at full speed, then estimates how it would load over a slower connection. Lantern handles that simulation. Its calculations were updated in version 13.6.0: the changes affect which requests are treated as render-blocking and how the time at which the largest element appears is determined on some pages.

In Lighthouse 13.5, whether a request blocked rendering was determined by its loading priority. Lantern now uses Chrome’s own render-blocking flag. If that flag is absent, it still falls back to priority. This changes the inputs to the simulation: two tests of the same page in different Lighthouse versions cannot be treated as entirely equivalent measurements.

The second change applies when the LCP element is text displayed in the same frame as the first content. Its appearance time is now calculated using the same page requests and tasks used for FCP. The previous calculation could include every request made before LCP. As a result, the simulation sometimes placed LCP noticeably later than FCP for the same text.

Developer Ivan Akulov demonstrated this on a test page built in Framer: above the fold, there was only text, which determined both FCP and LCP. Preloading JavaScript modules improved the real page’s time to interactive but made its calculated Lighthouse LCP worse. The issue was fixed as part of the Lantern update.

Pollard believes the new simulation is more likely to affect sites that rely heavily on JavaScript. One developer reported better results for single-page applications, citing react.dev as an example. Pollard clarified that the SPA format itself is not the reason. You can also read about field metrics for navigation within SPAs, but that is a different way of assessing user experience and is unrelated to the Lighthouse simulation change.

Why a lab score isn’t the same as visitor experience

The performance score in PageSpeed Insights comes from a Lighthouse lab test: page loading is simulated for one device under fixed network conditions. This lets you test a page in a defined environment, but a higher score alone does not prove that real visitors now see it load faster.

The same PageSpeed Insights report also includes CrUX field data. It is based on real visits over the past 28 days, so it is fundamentally different from the result of a single simulation. If the lab score rises after the update, it is worth checking that part of the report too. The Lighthouse version used for the lab result is shown next to the emulated device.

Lighthouse 13.6.0 also includes a change unrelated to speed scoring. The Agentic Resource Discovery audit now recognizes the new manifest filename, ard.json: previously, the audit might not detect a file with that name unless robots.txt contained an Agentmap line. This fix concerns resource discovery checks, not the calculation of loading metrics.

What to check before reporting an improvement

If a chart of lab scores suddenly jumps, first compare the Lighthouse versions used in reports from before and after the jump. Results obtained before and after the move to 13.6 are calculated under different rules. If work was done on the site during the same period, the score alone cannot tell you how much of the increase came from those changes and how much came from the test update.

Site owners should separate two questions: did the simulation score change, and did the visitor experience change? For marketers, this helps avoid attributing the entire rise in a lab metric to work done on the site. For developers, it means not judging the effect of an optimization by estimated FCP and LCP alone, but comparing reports from the same Lighthouse version and checking field data.

The change may appear in other tools as well: the Lighthouse team expects version 13.6 to be included in Chrome 159 DevTools. As of November 2, Chrome 159 was in the desktop Dev channel and Canary; its stable release was scheduled for December 8. Work on the simulation is not finished, however: an issue remains open on Lighthouse’s GitHub concerning cases where Lantern’s estimated LCP is significantly higher than the measured value, including on large JavaScript sites and SPAs.

My take

I wouldn’t treat a higher PageSpeed Insights score as proof that an online store has improved without checking the Lighthouse version and visitor data. When it comes to placing an order, what matters more to me is how the site works on a phone and whether shoppers drop off between the cart and payment. A lab test helps identify problems, but it cannot answer those questions on its own.

Sources

Where the news comes from. The text is a retelling in the author’s own words; the facts come from the source, the opinion is the author’s.

1. PageSpeed Insights Scores May Rise After Lighthouse Update via @sejournal, @MattGSouthern www.searchenginejournal.com

About the author

Rūta Jankauskaitė

natural cosmetics store owner

I sell cosmetics online and see the website first and foremost as part of the ordering process. Loading speed, clear delivery terms, and making sure customers don’t get lost between the cart and payment are all important.

All posts by the author