Large LCP jump recently
We’ve noticed a jump in our LCP over the past two days. While number of loads is lower, the jump in LCP is large, and the sample size is large enough to be meaningful.
There have been no deployments or code changes on our end during this period.
From our LCP logs, we can see that some loads are taking much longer than usual, followed by a subsequent fast load a minute or two later for the same store. This seems to suggest a possible pattern where the initial install or cold start is delayed, while follow-up loads are normal.
I've attached screenshots showing these patterns - same store, one very slow load followed by a normal one.
For context:
• We display static content during the initial install to optimise LCP (this setup has been stable for a year or two).
• The same static content is shown while loadingAuth to avoid blocking visual load while Gadget is loading too.
Could you please confirm if there have been any recent changes to Gadget’s cold start times or install process that might explain this?
Or is there anything in the install or load flow that could be contributing to slower initial responses?
This part of the process is mostly a black box for us - it’s one of the reasons we rely on Gadget, so I'm unsure what else to check internally.
There have been no deployments or code changes on our end during this period.
From our LCP logs, we can see that some loads are taking much longer than usual, followed by a subsequent fast load a minute or two later for the same store. This seems to suggest a possible pattern where the initial install or cold start is delayed, while follow-up loads are normal.
I've attached screenshots showing these patterns - same store, one very slow load followed by a normal one.
For context:
• We display static content during the initial install to optimise LCP (this setup has been stable for a year or two).
• The same static content is shown while loadingAuth to avoid blocking visual load while Gadget is loading too.
Could you please confirm if there have been any recent changes to Gadget’s cold start times or install process that might explain this?
Or is there anything in the install or load flow that could be contributing to slower initial responses?
This part of the process is mostly a black box for us - it’s one of the reasons we rely on Gadget, so I'm unsure what else to check internally.
