Hi everyone,
Since October 27, we have noticed a sudden spike in LCP (Largest Contentful Paint) scores across both of our Shopify apps.
We haven’t deployed any new changes recently, and this issue appeared unexpectedly. Based on the last 7 days of data, the LCP scores have increased significantly, with one app reaching approximately 9 seconds in the latest data points.
Has anyone else experienced a similar issue since October 27?
Could this be related to a recent Shopify-side change, or might there be another possible cause? Any insights or suggestions on how to investigate this would be greatly appreciated.
I’m attaching the LCP graphs for both apps covering the last 7 days for reference.
Thanks in advance!
No deploy plus a jump on a fixed date usually means the change came from outside your code. Split the 9s into its parts before changing anything. Open the app inside the admin, run a Chrome Performance trace with 4x CPU throttling, and check how long TTFB, resource load and render each take. A high TTFB points to your host: cold starts, a sleeping dyno, or an OAuth/session check on the first request. A long resource load usually means the App Bridge script is no longer loading first. It should be the first synchronous script in <head>, from cdn.shopify.com, with no async or defer. If most of the time is render delay, look at what your LCP element is. It’s often a large table or image that waits on a client fetch. Render a lightweight Polaris skeleton or server-rendered shell first, then fetch the data. Also compare by route in the Dev Dashboard’s web vitals: if only one page spiked, the problem is that page’s data call. If you want, I can take a closer look — happy to help by DM.
LCP in Shopify has been broken since release.
It’s not an accurate measure of customer experience, if you want to make it lower, just add a big square on top of your app while it loads, here’s an example of what pagefly does:
I am also having this issue, I thought it was because I upgraded to Polaris 2.0 but other apps are also showing the same
Seeing the same here since October 27, no changes from my side.
