Communicating During a Mass Incident: Building a Crisis Flow

On a normal day messages arrive one at a time. On an incident day — the carrier halts deliveries, a supplier misses a shipment, payments fail, a batch is recalled — the same question arrives hundreds of times at once. Without a plan, agents slow down retyping the same sentence while unanswered customers grow angrier.

What a Crisis Flow Is

A crisis flow is a temporary communication mode prepared in advance and switched on with one action. While active, the assistant knows the situation: it recognises affected orders, delivers the approved statement, shares the expected timeline and avoids unnecessary escalation. When the incident closes, the mode is switched off.

Five Things to Prepare Beforehand

• A rule for identifying the affected group: which orders, region, date range or batch.
• A three-stage message set — first announcement, interim update, closing notice — all written and approved in advance.
• Escalation thresholds: compensation demands, health risk, legal language, press enquiries.
• Authority: who switches the mode on, and who switches it off.
• Records: what was said to whom and when, for reconciliation afterwards.

Early and Incomplete Beats Late and Perfect

Waiting until everything is confirmed is the common mistake. Customers notice the problem before you announce it; what they want first is confirmation that you know. A good first message has four sentences: what happened, who is affected, what you are doing, and when you will update again. That last one measurably reduces repeat questions.

What Helps and What Hurts

In recalls involving safety, automation should deliver the notice and the instructions only; every follow-up question belongs to a person.

The Template Reality

Most affected customers have not messaged you in the last 24 hours, so free-form replies are not available. The crisis announcement has to exist as an approved template before the incident — waiting for template review during a crisis means losing the channel. One template with variables covers most scenarios.

Carrying the Volume

Incident-day volume is a multiple of normal. Know your messaging limits and team capacity in advance. The difference from a campaign day is simple: you schedule campaigns, you do not schedule incidents.

After It Closes

• Send a closing message: resolved, what was done, how any compensation works.
• Switch the mode off — a stale crisis message after resolution damages trust.
• Review the logs: most frequent questions, which answers worked, where escalation was unavoidable.
• Update the message set using the real sentences from this incident.
• Track affected customers for four weeks; quiet churn outnumbers loud complaints.

With BozLat AI

Crisis mode is bound to the affected order set: the assistant recognises those customers, delivers the pre-approved notice, tracks the promised update time and escalates conversations that cross defined thresholds. See pricing.