Case studybrentmovson.com

We shipped a sub-second edge for Brentmovson.

A premium fitness brand running a heavy Express API behind an unoptimized front end. We re-architected the data layer, streamed the critical render path, and rebuilt the image pipeline until median LCP fell below the one-second line on real hardware.

−68%median LCP

Engagement snapshot

Median LCP: −68%
LCP, real hardware: <1s
Lighthouse perf pts: +34
API round trip: 1

Industry / region

Fitness & Online Coaching — United Kingdom

Stack

Express APIEdge streamingSharp image pipelineNext.js

The challenge

Brentmovson's booking and coaching flows were powered by a monolithic Express API issuing N+1 queries on every page, with an unoptimized front end loading full-resolution hero imagery up front. Largest Contentful Paint sat well past three seconds on mid-tier mobile — slow enough to bleed both rankings and sign-ups.

The refactor

Data-layer rebuild

We replaced the N+1 Express queries with batched, cached resolvers so the API returned everything the page needed in a single round trip — collapsing time-to-first-byte before a single pixel rendered.

Critical-path streaming

The above-the-fold shell now streams immediately while heavier, below-the-fold data resolves inside Suspense boundaries. First paint no longer waits on the slowest query.

Image pipeline

Hero and gallery media were rebuilt as responsive AVIF/WebP with width-aware srcsets, explicit dimensions to kill layout shift, and priority hints on the LCP element only.

Implementation

app/page.tsxTSX

1// BEFORE: one blocking fetch stalled the entire render.
2// AFTER: stream the shell, defer the heavy data, ship real images.
3import { Suspense } from "react";
4import Image from "next/image";
5 
6export default function Page() {
7  return (
8    <>
9      <Hero />                        {/* paints immediately */}
10      <Suspense fallback={<StatsSkeleton />}>
11        <Stats />                     {/* streams when ready */}
12      </Suspense>
13    </>
14  );
15}
16 
17function HeroImage() {
18  return (
19    <Image
20      src="/hero.avif"
21      alt="Coaching session"
22      width={1280}
23      height={720}
24      priority
25      sizes="100vw"
26    />
27  );
28}

Streaming the shell behind a Suspense boundary while the LCP image ships as a sized, prioritized AVIF — the core of the sub-second rebuild.

The outcome

Median LCP fell 68% and crossed below the one-second line on real mid-tier hardware, with a 34-point Lighthouse performance gain. Faster pages compounded into higher organic visibility and a measurable lift in completed bookings.

More case studies

minded.es+1,420%pages indexed / mo nettleworth.com+1,883%non-US organic sessions coyotekits.com0msstate latency remmd.com+240%checkout conversion lifemeasure.io99.9%data accuracy PowerPulse Malta3.2xqualified lead lift Phil Hoffmann Travel170%organic growth Vibelab1modular platform model

Want numbers like these?

Start with a free 40-point technical audit — live crawl, instant health score, and the exact fix list. No login.

Run the free audit