Case study · Kantam Platform (name changed)
Cutting a marketplace TTFB from 1.9s to 210ms
A three-year-old Astro marketplace had drifted slow. Profiling, not guesswork, found the four real causes.
Client: Kantam Platform (name changed)Role: Performance ConsultantTimeline: 6 weeks
1.9s → 210ms Listing TTFB
95%+ Edge cache hit rate
3 Regressions caught by budgets
The challenge
Every product listing page took nearly two seconds to first byte. Traffic had tripled over two years and the team had bolted on features faster than they had revisited the architecture. Previous attempts at caching had made the site stale in ways the business noticed — the trust was to fix the causes, not hide them.
The solution
Profiling showed four compounding causes: a per-request CMS round trip that could be edge-cached with proper revalidation, unbounded image transforms on hot paths, a waterfalled third-party script, and N+1 queries in the related-items block. I fixed each in order of risk, added regression budgets to CI so the fixes stay fixed, and handed the team the profiling workflow.
The outcome
TTFB on listing pages fell from roughly 1.9s to about 210ms, cache hit rates now sit above 95%, and the CI budgets have caught three would-be regressions since handover.
“Abdullah fixed causes, not symptoms — and left the team able to keep it fast without him.”
astroperformanceedge