How to choose an on-call tool

Dozens of tools schedule on-call and escalate alerts. Most handle the core job. Differences lie in notification methods, pricing, and extra features. Review these questions before comparing feature lists.

1. How must people be notified?

This question rules out the most tools.

• Phone calls and SMS wake people reliably. Not every tool offers them. Some charge per message. Self-hosted tools need your own provider such as Twilio.
• Mobile push needs an app. Check it can override silent or do-not-disturb mode, if that matters to you.
• Chat (Slack, Microsoft Teams) is good in working hours and poor at 3am.
• Email is a record, not a page.

If you need people woken reliably at night, filter for phone and SMS first.

2. How do your schedules look?

Check the tool expresses your real rota, not a simplified one:

• Rotation length and handover time, per time zone.
• Several layers, such as weekday and weekend, or follow the sun.
• Overrides and swaps that people can make themselves.
• Restrictions, such as "only nights" or "only weekdays".

3. Does it take your alerts?

List every alert source you use today: metrics, uptime checks, logs, error tracking, cloud provider alarms, heartbeats. For each, check whether the tool has a native integration or whether you need a generic webhook. Native integrations usually map severity, resolve automatically and deduplicate correctly. A generic webhook may need work for each.

4. How is it priced?

• Per user is the most common model. Count everyone who needs an account, not only people on the rota. Some tools charge only for responders and let stakeholders view for free.
• Per team or organisation is flat and predictable.
• Add-ons: phone and SMS credits, status pages, incident management and advanced reporting are often priced separately.
• Free plans vary a lot: some cap users, some cap alerts, some leave out phone calls. The directory marks tools with a free plan.

Always check the vendor's current pricing page; plans change often.

5. Hosted or self-hosted?

Self-hosting gives you control over data and cost, and works in isolated networks. It also means the thing that tells you your systems are down runs on your systems. If you self-host, run it somewhere independent of what it monitors, and monitor it from outside. Filter for self-hosted on-call tools.

6. How much else do you want in one tool?

Some tools do on-call only. Others bundle uptime monitoring, status pages and incident management. A bundle means fewer vendors and one bill. Separate tools let you replace one part without the others. Neither is wrong. Decide which you prefer before comparing prices, because bundle prices and single-tool prices are not comparable.

7. What does leaving cost?

• Can you export schedules, policies, and incident history?
• Does it accept alerts in the format of the tool you use now? Several tools accept PagerDuty- or Opsgenie-compatible payloads, which lets you switch by changing a URL.
• Is the vendor stable? Check for end-of-sale announcements. Opsgenie stopped new sales on 2025-06-04 and shuts down on 2027-04-05, and the Grafana OnCall open source repository was archived on 2026-03-24.

Both are marked in the on-call directory.

A short trial plan

1. Recreate one real schedule and escalation policy.
2. Connect one real alert source.
3. Trigger a test alert at night with the phone on its usual settings and check it woke the right person.
4. Let it go unacknowledged and check it reached level 2 on time.
5. Make an override from a phone.
6. Look at the bill for your real headcount, including add-ons.

Compare the tools in the on-call and alert escalation directory.