Social media crisis monitoring that warns you in time
Social media crisis monitoring fails when it waits for a volume spike. The earlier signals are the reply ratio, unfamiliar accounts and what is being reposted.

By the time a crisis is obvious, the useful window has closed. Social media crisis monitoring is about the hour before it was obvious.
That is a measurement problem, and it is solvable.
What an early signal actually looks like
Not a spike in volume. Volume spikes late, after the thing has already spread. The earlier signals are a change in the ratio of replies to original posts, a sudden rise in accounts mentioning you that never had before, and a shift in which posts are getting reposted.
Any of those can move an hour or two before the volume chart does anything interesting.
Tell us what you need watched and we will define the baseline and the thresholds with you.
You need a baseline before you need alerts
An alert without a baseline fires constantly and gets muted within a week. That is the entire lifecycle of most monitoring setups. Two or three months of normal figures gives you a range: typical daily volume, typical reply share, typical concentration. A threshold set against that range means something. One picked out of the air does not.
Concentration decides the response
A spike carried by one large account and a spike carried by four hundred small ones look identical on a volume chart and need opposite responses. One is a conversation with a person. The other is a wave, and replying to individuals will not touch it.
Work out how many accounts make up half the volume. If the answer is one or two, you are dealing with a post, not a movement.
Reach matters more than mentions here
A thousand angry posts nobody saw is a smaller problem than fifty that reached a million people. Impressions tell you the size of the audience exposed. Without that figure you cannot tell whether something is a crisis or a squabble among people who already dislike you.
This is where counted figures matter rather than estimated ones, because the decision has consequences.
What to reconstruct afterwards
Once it is over, the retrospective is where the value is. Hour by hour volume and impressions, the first posts that moved, which account pushed it past the point of no return, and how long the decay took. That timeline is what makes the next playbook better than the last one.
The archive makes this possible after the fact, so a crisis nobody was tracking can still be reconstructed properly.
Making it usable by people who are not analysts
During a crisis nobody reads a dashboard. What works is a short note: how big, how fast, who is carrying it, whether it is still accelerating. Four sentences and a number for each. Everything else waits for the retrospective.
Everything we gather stays within the Twitter compliance rules. Tell us what you are exposed to and we will set the measurement up.