I believe the tests run from web.dev or PageSpeed (not devtools) might use different locations which can cause different results. @benschwarz will have more information, he’s done copious amount of research on this :)
Jeremy KeithJeremy Keith@rem
It’s actually devtools that I get the greater variation. But I’ve also put that down to background noise (ie. my browser/cpu doing something else that slows down during the perf check) causing small variances. But that’s a gut based guess.
Jeremy KeithJeremy KeithBen Schwarzfantastic ms.
Also thanks to you, I’ve got an utterly convoluted OTT service worker! :) Though, I’ve managed to get the titles, last visit, urls and publish dates of offline pages without the use of local browser storage 💪
Jeremy KeithJeremy Keith