Sveltekit

Function Duration dominates server-rendered routes (every +page.server.ts load + +server.ts POST handler runs as a function). Edge Requests grow with API surface (+server.ts and form actions). ISR is supported via the adapter; when enabled, it converts to a cache_result HIT after first render. Image Optimization is rarely a SvelteKit-specific lever (it’s the same Vercel image service Next.js uses).

Priority patterns

1. Adapter ISR for cacheable content. Routes that don’t depend on per-request data are still served as functions by default. The adapter accepts an isr: { expiration: 60 } option per route (set in +page.server.ts via export const config). This converts function invocations to cache hits. Cite https://kit.svelte.dev/docs/adapter-vercel + https://vercel.com/docs/incremental-static-regeneration.
2. Prerender what’s static. export const prerender = true in +page.server.ts or +page.ts moves a route from function to CDN. Cite https://kit.svelte.dev/docs/page-options.
3. Parallel load fetches. A load function with multiple sequential await fetch(...) calls leaves wall-clock time on the table — wrap them in Promise.all (or return promises directly from load, which SvelteKit streams). Cite https://kit.svelte.dev/docs/load.
4. Move per-request work to +server.ts action handlers and run them via fetch from the client. Reduces SSR cost when only a slice of the page actually needs server data on every request.
5. hooks.server.ts matcher hygiene. Like Next.js middleware, the handle hook intercepts every request unless filtered. Heavy handle code multiplies cost by request volume. Move work into the specific route’s load when only that route needs it.
6. Adapter runtime + region config. Single-region default; if the project’s users skew to a different region, set regions: [...] on the adapter to reduce TTFB by 100-300ms.

Frequent gotchas

• Per-route SSR when prerender would do. Marketing pages, docs, blog posts often end up as functions because nobody added prerender = true. The scanner flags these.
• +layout.server.ts data fetches blocking every child route. Auth-check + user-load in a layout makes EVERY function invocation wait on those queries — even routes that don’t read user. Push user-load into the routes that need it.
• Adapter version drift. adapter-vercel@5 adds new options (ISR, split). adapter-vercel@3 doesn’t. The recommender must check the installed version before suggesting isr: ....
• fetch calls in load to your own SvelteKit routes. SvelteKit optimizes these into direct module calls during SSR, but only if the URL is relative. A hardcoded https://your-domain.tld/api/... defeats this optimization.
• No connection pooling on serverless. Same as Next.js — Postgres without a pooler exhausts the database under load.

Cross-references

• https://kit.svelte.dev/docs/adapter-vercel — adapter config (ISR, regions, runtime)
• https://kit.svelte.dev/docs/page-options — prerender, ssr, csr
• https://kit.svelte.dev/docs/load — parallel fetches in load
• https://kit.svelte.dev/docs/routing — file conventions
• https://kit.svelte.dev/docs/hooks — handle / handleFetch
• https://kit.svelte.dev/docs/form-actions — server-side form handling
• https://kit.svelte.dev/docs/state-management — request-scoped state
• https://vercel.com/docs/incremental-static-regeneration — ISR on Vercel
• https://vercel.com/docs/fluid-compute — Fluid Compute (framework-agnostic)