Vercel's dashboard says "Cron Jobs." The vercel.json config accepts a cron expression. Deploy, and the schedule appears in your project settings. Everything looks production-ready until you check the fine print: on the Hobby plan, your jobs fire once per day, with up to 59 minutes of scheduling drift, and the function gets 10 seconds before Vercel kills it. That's not a cron system. That's a daily reminder with a shrug for timing. If you've been searching for a Vercel cron alternative, the limitations are real.
If you're evaluating serverless scheduling more broadly across AWS, Google Cloud, and other platforms, our serverless cron jobs guide covers that ground. This page focuses specifically on Vercel's built-in cron, where it breaks, and what to do about it.
What Vercel cron actually gives you (and where it stops)#
Vercel cron is built on Amazon EventBridge Scheduler under the hood. You define a path and a schedule in vercel.json, deploy, and Vercel sends an HTTP GET request to that API route on the production deployment — simple enough, but once you check which plan you're on, the limitations stack up fast.
Here's the real breakdown:
| Capability | Hobby (free) | Pro ($20/mo per seat) | Enterprise |
|---|---|---|---|
| Jobs per project | 100 | 100 | 100 |
| Minimum frequency | Once per day | Once per minute | Once per minute |
| Scheduling precision | ±59 minutes | Per-minute | Per-minute |
| Function timeout (default) | 10 seconds | 60 seconds | 60 seconds |
| Function timeout (max) | 300s (5 min) | 800s (~13 min) | 900s (15 min) |
| Log retention | 1 hour | 1 day | 3 days |
| Timezone support | UTC only | UTC only | UTC only |
| Retry on failure | None | None | None |
| Failure alerts | None | None | None |
That last column is the one worth staring at. Even on Enterprise, there are no retries and no failure alerts. Vercel won't notify you if a cron job fails, times out, or simply doesn't fire. The official docs say it plainly: "Vercel will not retry an invocation if a cron job fails."
The Hobby plan tax: paying $20/mo for basic scheduling#
The hard way — that's how most developers discover Vercel cron's limits. You build a Next.js app, add a /api/cleanup route that should run every hour, write the cron expression, deploy. The deploy succeeds. The job never fires more than once a day.
The Hobby plan caps frequency at daily. Not hourly. Not every fifteen minutes. Daily. And even that daily run has ±59 minutes of drift, so a job scheduled for 0 8 * * * (8 AM UTC) can fire anywhere between 8:00 and 8:59. If your job depends on timing, you're already in trouble.
Want minute-level scheduling? That's the Pro plan at $20 per month per team member. The per-seat pricing model means the cost scales with your team, not with the number of jobs you run — the opposite of how dedicated scheduling services price.
Vercel Pro costs $20/month per seat. A 5-dev team pays $100/month just to unlock minute-level cron scheduling. External cron services typically cost $0-7/month total for the same capability.
Why does your Vercel cron job time out?#
Two things work against you here: cold starts and tight timeout limits.
Vercel cron routes execute as standard serverless functions. If the function container is cold (and it will be, because cron jobs don't generate the steady traffic that keeps containers warm), you burn 1-3 seconds on initialization before your actual job logic runs. On the Hobby plan's 10-second default timeout, that leaves 7-9 seconds for your actual work. Try running a database migration or an API sync in that window.
The Pro plan bumps the default to 60 seconds (extendable via maxDuration), which helps. But the fundamental architecture remains: your cron job is a serverless function invoked by an HTTP GET, subject to the same cold-start lottery as any other request. There's no timeout strategy that fixes the underlying constraint. If your job needs 90 seconds reliably, you're fighting the platform.
No alerts, no retries, no execution history#
This is where Vercel cron falls apart for anything you'd run in production.
No retries. If a transient network error prevents the scheduled request from reaching your function, Vercel won't try again. The official docs acknowledge this: "a transient network error can prevent a scheduled request from reaching your function, and in that case no runtime log is created." Your job just didn't run, nobody knows.
No failure alerts. Your cron route throws an unhandled exception? Times out? Returns a 500? Vercel won't send you a Slack message, an email, nothing. You find out when a user reports stale data, or when you manually check the dashboard three days later. The Vercel community forums are full of threads titled "Cron Jobs Not Executing" from developers who only discovered the problem after hours or days of missed runs.
Minimal execution logs. Hobby plan retains logs for one hour. Pro gives you one day. If your cron job failed at 2 AM and you check at 9 AM, the logs are already gone on Hobby. You're debugging blind.
I've seen a postmortem where four cron jobs on a Vercel Pro project silently stopped firing for six hours. A stray duplicate path in vercel.json caused the cron registration to enter a paused state. No alert. No error in the dashboard. Just six hours of missed database syncs discovered by accident.
Common Vercel cron failures developers report
Jobs show as configured in Settings but never appear in execution logs. Manual
When Vercel cron is the right choice#
None of this means Vercel cron is useless. It works well in a narrow lane:
- Hobby projects with daily tasks. A personal site that rebuilds its RSS feed once a day? Vercel cron is fine. The timing drift doesn't matter, the timeout is probably enough, and you're paying nothing.
- Pro teams with simple internal triggers. If you're already on Vercel Pro for other reasons and your cron jobs are lightweight API route calls within the same project (cache invalidation, a quick database cleanup), the built-in cron avoids adding another service.
- Prototyping. Get the scheduling pattern right in development, then move to a proper service before production. Less setup friction than configuring an external tool for throwaway code.
Vercel cron works when timing doesn
What to look for in a Vercel cron alternative#
An external cron service takes a fundamentally different approach. Instead of coupling your schedule to a deployment config file, you configure a job in a separate dashboard (or via API) that sends HTTP requests to any public URL on whatever schedule you define. Your Vercel API routes still run the logic. The trigger just comes from outside.
What this gives you that Vercel cron can't:
- Retries with backoff. If your endpoint returns a 500, the service retries automatically (typically 3-5 times with increasing delays) instead of silently dropping the execution.
- Failure alerts. Missed a heartbeat? Response code isn't 200? You get notified through Slack, Telegram, email, or whatever channels you've configured. This alone solves the "found out three days later" problem.
- Full execution history. Every invocation logged with status code, response time, response body. Not for one hour. Not for one day. Persistent.
- IANA timezone support. Schedule "9 AM Berlin time" and it handles DST transitions without you redeploying twice a year.
- No deployment coupling. Change your schedule, add a new job, pause and resume. No
git pushrequired. - Platform-agnostic. Works with Vercel, Railway, Render, a VPS, WordPress — anything with a public URL, no server required. Move your hosting tomorrow and keep every schedule intact.
The trade-off is adding an external dependency — if the cron service itself goes down your jobs stop firing — but dedicated scheduling services typically offer better uptime guarantees than a "best-effort" feature bolted onto a hosting platform, because scheduling is their entire product.
Vercel cron vs external cron: side-by-side#
Beyond the plan-level limits already covered, a Vercel cron alternative differs in several architectural ways:
| Capability | Vercel Cron (Pro) | External Cloud Cron |
|---|---|---|
| HTTP methods | GET only | GET, POST, PUT, PATCH, DELETE |
| Custom headers / auth | CRON_SECRET only | Any header, HTTP Basic Auth |
| Response body capture | No | Yes |
| Schedule changes | Requires redeploy | Dashboard or API, instant |
| Call external URLs | Own project only | Any public URL |
| Concurrent execution protection | Manual (Redis locks) | Built-in overlap prevention |
| Cost (5-person team, minute-level cron) | $100/mo | $0-19/mo |
The cost line deserves a closer look. Vercel Pro is per-seat, and you can't buy it for just one developer — the whole team upgrades or nobody does. External cron services charge per workspace, not per person, so a growing team doesn't multiply the bill. The savings compound fast once you factor in the monitoring, retries, and alerting you'd otherwise need to build yourself.
Moving your schedules off vercel.json#
The migration is straightforward because Vercel cron jobs are already HTTP endpoints. You're not rewriting business logic. You're changing who sends the request.
Here's a typical vercel.json cron config before the migration:
{
"crons": [
{
"path": "/api/cleanup",
"schedule": "0 3 * * *"
},
{
"path": "/api/sync-orders",
"schedule": "*/15 * * * *"
}
]
}
Both routes stay. The crons block goes.
Step 1: Keep your API routes. Your /api/cleanup, /api/sync-orders, whatever they are — they're regular serverless functions and work the same whether Vercel or an external service triggers them.
Step 2: Secure the endpoint. Replace the CRON_SECRET bearer token check with whatever auth your external service sends. Most services let you configure custom headers, so you can keep the same Authorization header pattern.
Step 3: Set up the external schedule. In your cloud cron dashboard, create a new job pointing at https://your-app.vercel.app/api/cleanup with your cron expression. Set the timezone, configure retries, pick your alert channels.
Step 4: Remove the cron entry from vercel.json. Delete the crons array and redeploy. The API routes themselves stay. Only the trigger source changes.
Step 5: Verify. Watch the first few executions in your external service's execution log. Confirm the status code, response time, and response body look right. Set up a dead man's switch alert for missed runs if the job is critical.
Fifteen minutes per job. That's the migration. No code changes.
You can run both systems in parallel during migration. Keep Vercel cron active while the external service warms up, then remove the vercel.json entry once you
The hybrid approach: Vercel cron + external monitoring#
Not everyone needs to move scheduling entirely. If your Vercel cron setup works but the lack of alerting bothers you (and it should), there's a middle path: keep Vercel cron as the trigger, add heartbeat monitoring on top.
The pattern: at the end of your cron route, send an HTTP request to a monitoring endpoint (a ping URL). If the ping stops arriving on schedule, the monitoring service alerts you. You get failure detection without changing your scheduling infrastructure. It's the same dead man's switch pattern used for traditional server cron jobs, applied to serverless.
This works well for teams that are happy with Vercel Pro's scheduling but can't tolerate the "nobody told me it failed" problem. The monitoring layer catches silent failures, triggers incidents, and gives you the execution history that Vercel's one-day log retention doesn't.
Frequently asked questions#
Can Vercel cron jobs run more often than once per day on the Hobby plan?
No. Daily only, with up to 59 minutes of drift. You need Vercel Pro ($20/month per team member) or an external cron service for anything more frequent.
What is the maximum timeout for a Vercel cron job?
The default timeout is 10 seconds on Hobby and 60 seconds on Pro. You can extend it with the maxDuration config: up to 300 seconds (5 minutes) on Hobby, 800 seconds (~13 minutes) on Pro. Jobs needing longer than that require a different execution model or an external trigger paired with a background queue.
Why is my Vercel cron job not running?
Common causes: the cron expression is more frequent than daily on a Hobby plan (silently fails at deploy), the function hits a cold start and times out before completing, or the CRON_SECRET environment variable isn't set in production so your auth middleware blocks every invocation. Check the Vercel function logs within the first hour (Hobby) or first day (Pro) before they're purged.
Does Vercel cron support timezone scheduling?
No. All Vercel cron expressions run in UTC. If you need a job to fire at 9 AM Berlin time (which shifts between UTC+1 and UTC+2 across DST transitions), you'd need to manually adjust the expression twice a year and redeploy, or use an external service with IANA timezone support.
Can I use an external cron service to trigger Vercel serverless functions?
Yes. Any service that sends HTTP requests on a schedule can call your Vercel API routes. Your functions don't care who sends the request — Vercel, an external scheduler, a curl command from your laptop, it's all the same. Secure the endpoint with a shared secret header, point the external service at your production URL, and the function executes normally.
What happens if a Vercel cron job fails?
Nothing. Vercel doesn't retry the invocation, doesn't send an alert, and doesn't flag it in the dashboard. The job simply doesn't run, you wait for the next scheduled time. If the failure was caused by a transient error, no runtime log is created at all.
Is it worth paying for Vercel Pro just for cron jobs?
Probably not, unless you need other Pro features (deployment previews, analytics, team management). External cron services offer retries, alerts, timezone support, and POST requests for $0-19/month total regardless of team size. If you're already on Pro for other reasons, the cron is a free bonus — but upgrading solely for cron scheduling rarely makes financial sense.
How do I monitor Vercel cron jobs for failures?
Add a heartbeat ping at the end of your cron route. When the job completes successfully, it sends an HTTP request to a monitoring endpoint. If the ping stops arriving, you get alerted. This works whether you keep Vercel's built-in cron or switch to an external Vercel cron alternative.
Frequently asked#
No. Daily only, with up to 59 minutes of drift. You need Vercel Pro ($20/month per team member) or an external cron service for anything more frequent.
The default timeout is 10 seconds on Hobby and 60 seconds on Pro. You can extend it with the maxDuration config: up to 300 seconds (5 minutes) on Hobby, 800 seconds (~13 minutes) on Pro. Jobs needing longer than that require a different execution model or an external trigger paired with a background queue.
Common causes: the cron expression is more frequent than daily on a Hobby plan (silently fails at deploy), the function hits a cold start and times out before completing, or the CRON_SECRET environment variable isn't set in production so your auth middleware blocks every invocation. Check the Vercel function logs within the first hour (Hobby) or first day (Pro) before they're purged.
No. All Vercel cron expressions run in UTC. If you need a job to fire at 9 AM Berlin time (which shifts between UTC+1 and UTC+2 across DST transitions), you'd need to manually adjust the expression twice a year and redeploy, or use an external service with IANA timezone support.
Yes. Any service that sends HTTP requests on a schedule can call your Vercel API routes. Your functions don't care who sends the request. Secure the endpoint with a shared secret header, point the external service at your production URL, and the function executes normally.
Nothing. Vercel doesn't retry the invocation, doesn't send an alert, and doesn't flag it in the dashboard. The job simply doesn't run until the next scheduled time. If the failure was caused by a transient error, no runtime log is created at all.
Probably not, unless you need other Pro features (deployment previews, analytics, team management). External cron services offer retries, alerts, timezone support, and POST requests for $0-19/month total regardless of team size. If you're already on Pro for other reasons, the cron is a free bonus — but upgrading solely for cron scheduling rarely makes financial sense.
Add a heartbeat ping at the end of your cron route. When the job completes successfully, it sends an HTTP request to a monitoring endpoint. If the ping stops arriving, you get alerted. This works whether you keep Vercel's built-in cron or switch to an external cron service.