Monitor your SaaS on Vercel — health URL + status page

Vercel makes shipping a Next.js frontend trivial. Production still needs a health endpoint the world can ping and an external monitor that is not your browser tab. This walkthrough uses the App Router (app/api/health/route.ts), deploys to a Vercel URL, and registers a StillOnline check plus auto status page.

Generic steps live in Health check quickstart. Railway users: Railway guide.

1. Add `app/api/health/route.ts`

Keep the handler dynamic if you touch the database — static optimization can cache a stale “ok”.

import { NextResponse } from "next/server";

export const dynamic = "force-dynamic";

export async function GET() {
  let db: "ok" | "error" = "ok";

  try {
    // Optional: await prisma.$queryRaw`SELECT 1`;
    // Remove DB ping for pure liveness on Free-tier hobby DBs.
  } catch {
    db = "error";
  }

  const healthy = db === "ok";
  const status = healthy ? "ok" : "degraded";

  return NextResponse.json(
    {
      status,
      service: "my-saas",
      db,
      timestamp: new Date().toISOString(),
    },
    { status: healthy ? 200 : 503 },
  );
}

Trade-off: returning 503 when Postgres blips marks you down on StillOnline — correct for “API not usable,” noisy during migrations. For liveness only, return 200 and omit the DB block.

2. Deploy on Vercel

Connect the Git repo, set production branch.
Add env vars (DATABASE_URL, etc.) in the Vercel dashboard.
Note the production host: https://your-app.vercel.app or custom domain.

Verify:

curl -sS https://your-app.vercel.app/api/health

3. Avoid common Vercel footguns

Issue	Fix
Middleware blocks `/api/health`	Exclude path in `middleware.ts` matcher
Edge runtime without DB driver	Use Node runtime for DB health or skip DB
Deployment protection on previews	Monitor production URL only
Redirect `www` → apex	Use the final HTTPS URL in StillOnline

4. StillOnline check (five clicks)

Start free.
Create project + status slug.
HTTP check → https://your-app.vercel.app/api/health (or custom domain), expect 200, 5 min interval.
Share https://stillonline.tech/en/s/{slug} in footer / docs.
Optional Pro: MCP to recreate checks after domain changes.

Free: 1 URL, 24h history. Pro $9/mo: 10 URLs, private pages, API/MCP, 90d history — Pricing.

StillOnline does not deploy to Vercel and does not read Vercel logs. It only HTTP-gets your URL.

5. When Vercel is up but you are “down”

Serverless cold start — rare timeouts; increase monitor timeout tolerance or warm critical paths.
Region-only outage — single-region Vercel + single monitor = blind spot; advanced teams add a second check from another vendor (comparison post).
Third-party API — health should reflect your SLA, not Stripe’s; document external deps on the status page manually via incidents.

FAQ

Server Components vs Route Handler?

Use a Route Handler under app/api/health/route.ts for a simple GET probe. Pages can SSR-cache misleading success.

Should I monitor the marketing page or /api/health?

Monitor the API if paying users depend on it. Marketing on Vercel can be up while API env is misconfigured.

Does @vercel/speed-insights replace uptime monitoring?

No. Speed Insights is performance sampling, not availability alerts.

Can I use the same repo pattern as StillOnline?

Our production app uses a similar JSON health with DB ping at api/health — see our open patterns in docs; your schema can be smaller.

Preview deployments?

StillOnline Free has one URL. Point it at production. Use Pro for extra URLs if you must watch staging.