LCP from 5 to 1.9 seconds, fixed during one call
A local SEO company with a brand new Next.js site, fast on desktop and slow on mobile.
A new site that passed on desktop and failed on mobile
gmbapi is a local SEO software company, and a good one. In July 2026 they dropped WordPress for a new site in React and Next.js on a headless CMS. Desktop met the Core Web Vitals. On mobile, LCP, FCP and INP did not, and their guess was unused JavaScript.
gmbapi, local SEO software
React, Next.js, headless CMS
Power hour: one call with their two developers, August 2026
Their developers, during the call
Death by a thousand cuts
There was no single big problem, and there almost never is. From the top of the page to the bottom, I walked them through the page timings, and we fixed things while we talked.
-
JavaScript in the right order
Scripts the first screen needs load first. The rest waits until the page is there.
-
Caching
Server response went from 3 seconds to 700 milliseconds. Every millisecond there comes off FCP and LCP before anything else is touched.
-
Resource scheduling
The hero image and the fonts go out early. Everything else gets out of their way.
-
A lot of small things
None of them fixes the site alone. Rewriting a 30 kB and a 0.5 kB stylesheet to 15 kB each saves 100 ms. A hero image that is no longer lazy loaded saves another 120 ms.
Their mobile visitors in CoreDash, per page and per release
After the call I gave them access to CoreDash, so they see their real mobile visitors per page and per release, and I checked their task list. The layout shifts are gone and the numbers above are from CrUX, Google's own field data.
An hour is enough when the developers are on the call and the fixes are small. Bigger things go in the notes, in the order to build them.
An hour with your developers on the call: €350
Bring your developer and your slowest page. We look at it together and fix what we can while we talk.
More results on my clients page.
