Multi-Channel Incident Alerting: How to Escalate From Push to SMS to a Phone Call

What's in this guide

• What push, SMS, voice, email and Slack or Teams are each good and bad at
• The default ladder, with a worked timeline from 03:00 to 03:15
• How to get through iOS Focus and Android Do Not Disturb at 3 a.m.
• Quiet hours and regional rules: who should not be woken, and when
• What to do when SMS or voice delivery fails
• What channels cost across on-call tools

This guide is about alerting your own on-call engineers when something breaks. If you are looking for an SMS or voice API to send notifications to your customers, that is a different category of tool (CPaaS providers such as Twilio or Vonage), and most of the advice below does not apply.

What each channel is good and bad at

No single channel is reliable on its own. Push needs data and an app that has not been force-closed. SMS needs a cell signal and a carrier that does not filter the message. A phone call needs someone who does not reflexively decline unknown numbers. Multi-channel alerting works because each channel covers the gaps of the one before it.

The pattern in that table is the reason for the order. Push is the cheapest interruption with the most information, so it goes first. Each later step is more intrusive and carries less detail, so it only fires if the earlier one did not work.

A note on Slack and Teams: posting an incident to a team channel is useful, but it is not an escalation step. A message to fifteen people at 3 a.m. is a message that nobody answers, because everyone assumes someone else will. Use chat for coordination, and use personal channels for the person who has to act. We cover the chat side in our guide to Slack incident management.

The default ladder: push, then SMS, then a phone call

Here is what a critical alert looks like at 3 a.m. with the default ladder. The engineer on call is Anna, and the secondary on call is Ben.

If Anna acknowledges at any point, the ladder stops. The SMS and the call are not sent, and Ben is never woken. That is the whole point: each extra channel only fires when the one before it failed.

Why 5 minutes between steps? Shorter than about 3 minutes and the SMS arrives while Anna is still unlocking her phone, which adds noise without adding reach. Longer than 10 minutes and a real outage runs unattended for half an hour before the backup hears about it. Five minutes per step, and 15 minutes before escalating to the next person, is the range most teams settle on for critical incidents.

"Escalate if they don't open it within 10 minutes." A common version of this question asks how to escalate from an in-app notification to SMS if the user does not open it. In incident alerting, the trigger should be acknowledgement, not whether a notification was opened or delivered. A push can be delivered and even opened by a sleepy thumb without anyone taking the incident. Only an explicit acknowledgement should stop the ladder.

How to set this up in All Quiet

In All Quiet, the escalation policy decides who is alerted and when the incident moves to the next person. Each responder decides how they are alerted, in their own notification settings: which channels are switched on, a delay per channel (for example push at once, SMS after 5 minutes, a call after 10 minutes), and which channels are used for which severity. That split means a team lead can own the policy while each engineer tunes the channels that actually wake them.

To get the timeline above:

1. In each responder's notification settings: push at 0 minutes, SMS at 5 minutes, phone call at 10 minutes, for critical incidents.
2. In the escalation policy: Tier 1 is the primary on-call schedule, Tier 2 is the secondary. Click "Then, doesn't auto-escalate" under Tier 1, set the "Escalation Cadence (min)" to 15 and turn on "Auto-escalate to next tier". The cadence counts from the first alert to Tier 1, not from the SMS or the call, so Ben is alerted at 03:15, after Anna's push at 03:00, SMS at 03:05 and call at 03:10.
3. For lower severities: switch SMS and calls off, so a warning sends a push or an email and waits for the morning.

For six complete policies you can copy, with step, who, channel and wait time, see our escalation policy examples. For the schedule behind Tier 1 and Tier 2, see the on-call rotation guide. For a product overview of channels, apps and integrations, see our page on multi-channel alerting.

Which tools can push alerts across email, SMS and voice at once?

Most dedicated on-call and incident alerting tools can send the same alert by email, SMS and voice at once, All Quiet included. The more useful question is whether they should. Sending every channel at the same second means four interruptions for one alert, and responders quickly learn to treat that as noise.

The better default is staged delivery: push first, then SMS, then a call, each a few minutes apart, stopping as soon as someone acknowledges. Reserve simultaneous delivery for the incidents where minutes matter most. In All Quiet, you get simultaneous delivery by setting every channel's delay to 0 for critical incidents, and staged delivery by giving each channel its own delay. Both run on every plan.

If what you need is to send email, SMS and voice to customers rather than to your own engineers, you are looking for a communications API instead of an on-call tool.

Critical alerts on iOS and Do Not Disturb on Android: getting through at 3 a.m.

The ladder only works if the phone actually makes a sound. Most engineers sleep with Focus or Do Not Disturb on, and by default that silences every app, including the one that is supposed to wake them.

On iPhone, Apple's Critical Alerts are the exception. According to Apple's developer documentation (checked 19 October 2026), a critical notification "bypasses the mute switch to play a sound" and is presented "even when Do Not Disturb is active". Apps need an entitlement approved by Apple to send them. All Quiet has that entitlement: after installing the iOS app, it walks you through allowing Critical Alerts. We wrote up the details in critical alerts on iOS.

On Android, the All Quiet app can override Do Not Disturb, so critical incidents still play a sound and vibrate while everything else stays quiet. The setup is covered in overriding Do Not Disturb on Android and in the alerting channels documentation.

Three things to check before your first night on call:

• Critical only. Use the Do Not Disturb override for critical incidents only. If warnings also break through, people turn the override off, and then nothing does.
• Allow calls and SMS from the alerting numbers. Save the numbers your on-call tool calls and texts from as a contact, and add that contact to the people allowed through Focus or Do Not Disturb. Otherwise steps 2 and 3 of the ladder can arrive silently. All Quiet calls and texts from fixed numbers, and the alerting channels documentation has a contact card with all of them.
• Send yourself a test. Trigger a test incident with the phone locked, on silent and in Do Not Disturb. If you do not hear it, neither will you at 3 a.m.

Quiet hours and regional rules: who should not be woken, and when

Quiet hours in incident alerting are not one setting. They are the result of three decisions: who is on call, which incidents are allowed to wake them, and what happens to everything else.

1. Only the person on call is alerted. The simplest quiet hours rule is the schedule itself. In All Quiet, notifications only go to team members who are on call, so everyone else is off duty by default. For teams in several regions, a follow-the-sun schedule hands over between regions, so each engineer is on call during their own daytime and nobody is woken at night for routine incidents. Our escalation policy examples include a two-region setup.

2. Only critical incidents escalate outside working hours. Severity is the filter that decides what is worth waking someone for. A critical incident runs the full ladder at any hour. A warning sends a push or an email and waits for the morning. In All Quiet, escalation tiers can be limited by severity and by time, and routing rules can mute, reroute, change the severity of or discard alerts by attribute, integration or time window. That is how you encode "nothing below critical pages anyone between 22:00 and 07:00" without relying on people to mute their phones. If you have not defined severities yet, start with our guide to incident severity levels.

3. Responders choose their own channels. Some engineers want a call for every critical incident; others want push only until the second step. Personal severity preferences let each responder decide which channels fire for which severity, inside the limits the team agrees on.

Regional rules. In some countries, contacting staff outside working hours is regulated. In Australia, for example, the right to disconnect lets employees refuse contact outside working hours unless the refusal is unreasonable. It has applied to employers with 15 or more employees since 26 August 2024, and to small businesses since 26 August 2025 (checked 19 October 2026). One factor in deciding what is reasonable is whether the employee is paid to be available. In practice, that makes formal on-call arrangements, with agreed hours and compensation, the safe way to wake people, and it makes alerting people who are not on call the thing to avoid. This is not legal advice; check the rules in each country where your responders work with your HR or legal team.

Can a multichannel platform manage regional quiet hours?

Yes, but not by itself. A multichannel alerting platform can enforce quiet hours once you have decided what they are: on-call schedules per region decide who can be woken, severity and time-window routing rules decide which alerts are allowed through at night, and personal channel preferences decide how each person is reached. What the platform cannot know is your local employment rules or on-call agreements. Those you encode in the schedules and rules. In All Quiet, schedules, escalation filters and routing rules are all available on every plan.

Failover: what happens when SMS or voice delivery fails

Channels fail more often than people expect. SMS can be delayed or filtered by carriers, particularly across borders. A call goes to voicemail because the phone is off, in airplane mode or out of signal. A push never arrives because the app was not set up on a new phone. Multi-channel alerting is your failover, as long as the ladder is built on acknowledgement rather than delivery.

• Acknowledgement stops the ladder, nothing else. A delivered SMS or a call that rang out is not a response. The next channel should fire unless someone explicitly acknowledged the incident.
• Voicemail is not an acknowledgement. A call that ends in voicemail should count as unanswered, so the escalation continues. In All Quiet, you acknowledge in the mobile app, the web app or Slack, so a call that rang out or went to voicemail does not stop the escalation.
• Mix data and cellular channels. Push travels over data or Wi-Fi. SMS and calls travel over the cellular network. Having both on the ladder means a dead mobile data plan or a hotel Wi-Fi without signal does not leave anyone unreachable.
• The final failover is another person. No channel works if the phone is in a drawer. That is why every ladder ends with escalation to a second responder, and why the last tier of an escalation policy should be a group or a manager, never "nothing".
• Test after every phone change. New phone, new SIM, new country: send a test incident and check all channels arrive.

For noise on the other side, alerts that fire every night and get acknowledged without action, see how to reduce alert fatigue. A ladder that rings too often is as dangerous as one that does not ring at all.

Which on-call tools send alerts by SMS, phone call, push, email and Slack or Teams without charging extra per channel?

Most paid on-call plans include SMS, phone calls, push and email, so "per channel" pricing is usually not where the cost difference is. The differences are in free tier caps, fair use limits, international coverage, and whether Slack or Teams is part of the plan you can afford.

Every All Quiet plan includes unlimited SMS, voice calls, push notifications and email, plus Slack and Microsoft Teams, with no per-channel or per-message charges. Mattermost is included as well. The Standard plan is $4.99 per user per month on the US price list, and EU-hosted accounts are billed 6,99 €, because the price follows the hosting region.

For comparison, as checked on 19 October 2026: PagerDuty's pricing page lists SMS, phone, push and email as included on its paid plans, starting with Professional at $25 per user per month billed monthly or $21 billed annually. Its Free plan includes 100 SMS or phone notifications per month and a basic chat experience, without Slack channel creation or messaging.

Sources and further reading

All product and competitor details checked on 19 October 2026.

• Alerting channels, All Quiet documentation. Per-user channels, delays, severity preferences, iOS Critical Alerts and Android Do Not Disturb override.
• Escalations, All Quiet documentation. Tiers, repeats, auto-escalation, severity and time filters.
• Pricing, All Quiet. Plans, prices and the alert channels included in each.
• UNNotificationInterruptionLevel.critical, Apple Developer Documentation.
• Incident Management pricing, PagerDuty. Plan prices and notification limits.
• Right to disconnect, Fair Work Ombudsman (Australia).