Skip to main content

Channels

WHAWIT pages on-call responders over seven channels, configurable per escalation level:

WhatsApp

Reaches people at night, worldwide, without a separate app to install.

SMS

The fallback that works when data does not. Requires a phone number on the user’s profile.

Phone call

Rings the handset and reads the page aloud. For the tier where a notification is not enough.

Slack

Incident messages with in-place actions, and /whawit in the channel. See the Slack guide.

Microsoft Teams

Incident cards with in-place actions. See the Teams guide.

Discord

For teams that run their operations there.

Email

Always available, no configuration required.
WhatsApp, SMS and phone calls need a phone number. They are delivered to the number on the recipient’s user profile. If a user has no phone number set, those channels are skipped for them — the escalation continues on the others. Fill in phone numbers before you rely on a policy that escalates to SMS.

Setting channels per level

Channels are a property of the escalation level, not of the policy. That lets urgency ramp up as the incident ages:
If a level does not specify channels, WHAWIT falls back to the recipient’s own notification preferences.

The phone call

A call is the loudest thing WHAWIT can do, and it behaves differently from the written channels in three ways worth knowing before you rely on it. It never travels alone. A tier that calls must also send SMS or WhatsApp, and the editor will not let you save one that does not. A spoken URL cannot be clicked, so the call carries no link — it points at the message that did. It goes out a minute after the text. If the responder reads the SMS and acknowledges inside that minute, the call is cancelled before it is placed. That is deliberate: the call exists for the page that went unread, not for the one that was already being handled. Pressing 1 on your verified phone acknowledges. The call asks the listener to press 1. On the number you verified in Preferences, that acknowledges every incident the call covers, as you, and the calls stop. On any other number it only confirms receipt: the calls stop and the confirmation is recorded against the page, but the incident still needs acknowledging from WHAWIT, Slack or Teams — a keypress proves somebody answered a number, which is equally true of a voicemail box or a forwarded line, and acknowledgement is the one action that stops a page reaching the next person. One call at a time per person. When several incidents page the same responder within a few minutes, they share one call that names each of them (“You have 2 incidents requiring acknowledgement. Incident 1: …”). Retries chase whichever are still unacknowledged, and one keypress covers them all. A call placed by hand from an incident is never folded into another. One incident per cause. While an incident for a monitor and service is still open, a new analysis of the same service is recorded on it as a recurrence — the timeline shows it and the severity rises if the recurrence outranks it — instead of opening and paging a second incident. After a day of quiet, the next analysis opens a fresh one. An unanswered call — no answer, busy, or picked up by an answering machine — is placed again a minute later, up to three calls in total. Acknowledging the incident stops the retries immediately.
Calling needs the same Twilio credentials SMS does. On a deployment without them the phone-call option is greyed out rather than silently doing nothing.

Calling the on-call by hand

You do not have to wait for a tier. Call on an incident’s action bar rings whoever is on call for that incident right now, reading the same page aloud. It does not move the incident to the next escalation level, and it goes through even on an acknowledged incident — you can see the state, so WHAWIT does not second-guess you.

What the page contains

The notification is not a bare “something broke”. It carries:
  • The severity and summary — what failed, and how badly.
  • What WHAWIT already found — the root-cause hypothesis from the investigation that ran before the page was sent, with the evidence behind it.
  • A suggested fix, where the investigation produced one.
  • An acknowledge link — one tap, and the escalation stops.
An example of what lands on a phone at 3 AM:
This is the difference worth internalising. A traditional paging tool routes the alert correctly and stops there. The engineer then starts the investigation. With WHAWIT the investigation has already run, so the engineer’s first decision is whether the conclusion is right — not where to begin looking.

Acknowledging

Acknowledging from the link in the notification stops the escalation immediately: no further levels fire, and no repeat runs. Until someone acknowledges, WHAWIT keeps working through the chain on its own.

Incident channels

Separately from paging individuals, WHAWIT can open a dedicated Slack or Microsoft Teams channel for the incident, assign responder roles, keep a running timeline, and draft the status updates and the postmortem. Paging gets the right person awake; the incident channel is where the response is coordinated once they are.

Next

Escalation policies

Where channels are configured, level by level.

On-call FAQ

Direct answers about coverage, migration and what is included.