In a crisis, the most important measure is whether people got accurate, usable information and whether the underlying harm was reduced. The number of press mentions may matter for reach, but it is not the outcome. Establish the incident timeline, affected audience, required action, and approved source of truth. For health, safety, legal, or security incidents, coordinate measurement with the responsible specialists.
Measure the response in stages
Speed and accuracy
Record when facts were confirmed, when each audience was notified, and when corrections were issued. Faster is not better if the statement is wrong.
Reach and access
Check whether affected groups actually received the update in a usable language and format, including people outside the usual email or social audience.
Understanding
Test whether people know the present risk, the action to take, and where to get help. Review recurring questions and misconceptions.
Protective action
Track the relevant behavior—product returns, account resets, use of a support line, or compliance with an official instruction—without confusing it with message exposure.
Resolution and trust
Record unresolved cases, repeat incidents, and later confidence, with a consistent method and a realistic time horizon.
A simple response log
For every public update, store its timestamp, approving owner, facts known, unresolved questions, channels, and correction history. Link operational measures to the communication timeline. If a data-breach notice leads to account resets, compare completed resets with the number of affected users who could access the instructions. If a recall notice is issued, coordinate return or remedy figures with the product-safety team rather than estimating them from web traffic.
The CDC's Crisis and Emergency Risk Communication resources emphasize clear, actionable communication during emergencies. AMEC's framework helps separate message outputs from audience understanding and behavior. Use both ideas without treating a communications dashboard as a substitute for incident command.
What not to claim
A good evaluation helps improve the next update while the response is still underway. It should make uncertainty visible to decision-makers, not produce a flattering retrospective score.
- Do not call sentiment recovery proof that the harm has ended.
- Do not count an email as received if addresses bounced or people could not access it.
- Do not imply a communications campaign alone caused a safety or financial outcome.
- Do not hide corrections because they temporarily worsen a dashboard.