Performance optimization, LCP under 1.5s, INP under 200ms, and we tell you what changed.
Performance work compounds in both directions. A well-tuned app stays fast as it grows; a neglected one accretes regressions until 'why is this slow' becomes a quarterly project. We tune the surfaces that matter, landing pages, the first authenticated screen, the largest queries, and instrument the rest so regressions surface before customers do.
What we build
Core Web Vitals tracked in CI
LCP, INP, CLS measured against a real-device profile in Lighthouse CI on every PR. Regressions fail the build with a delta against main. The targets aren't 'best-effort', they're enforced.
Postgres EXPLAIN as a routine, not a panic move
Slow queries get an EXPLAIN ANALYZE in the PR description. Indexes added against real plans. We don't add indexes speculatively; we don't leave them out optimistically. The query plan is part of the code review.
Caching strategy per surface
HTTP cache headers, CDN edge cache, Redis application cache, browser cache, each chosen against the data freshness and traffic shape of the surface. Marketing pages get aggressive CDN caching; product surfaces get carefully-scoped Redis caching with explicit invalidation.
Image pipeline that doesn't bottleneck the page
next/image (or equivalent) with proper srcsets, AVIF where supported, lazy-loading below the fold, blur placeholders. Image budget per page tracked in CI; oversized assets get caught at PR time.
Bundle size budgets per route
Per-route JavaScript budgets enforced via Lighthouse CI thresholds. New dependencies that push routes over budget have to justify themselves in the PR. Marketing surfaces stay particularly tight.
RUM as the production source of truth
Real-User Monitoring (Web Vitals + custom marks) reports the metric distribution from production traffic. Lab metrics are the ceiling; RUM is the floor. We tune against both.
Where this fits
Your homepage was fast when you launched and now LCP is 4 seconds and you don't know why.
Your dashboard takes eight seconds to load and the engineering team has been promising to look at it for two quarters.
Your product is on Lighthouse 50–60 across the board and the next sales pitch is going to a customer who measures these things.
Tech stack
- Lighthouse CI
- Web Vitals
- Postgres EXPLAIN
- Caching
Want this for your team?
30 minutes with a founder or senior engineer. We'll scope what you need and tell you straight whether Stacklane fits.
Book a Free CallRelated capabilities