Comments and crisis: the first hour decides

The article on online reputation names the switching threshold and stops there: several complaints about the same fact within hours, a pick-up by somebody outside, or a serious subject. This page picks up from that threshold.

What changes beyond it is not the intensity but the nature of the work. In ordinary mode you answer people one at a time. In crisis mode you address an audience, and every individual reply becomes an additional version of the story.

The first hour decides nearly everything, and it decides with very little information. That is the central difficulty of this subject: you have to write fast while knowing little, without asserting what cannot be verified and without appearing to say nothing.

This article sets out the sequence, the exact shape of the holding statement, the internal half nobody prepares, and the subjects that have to leave communications and go elsewhere. It also refuses two things we are nearly always asked for in the first hour.

The simplest rule in this article is also the most broken: one person writes, in one place, and everything else points at that place. No parallel replies, no version adapted per platform, no individual initiative.

The place is chosen on a single criterion: where the episode is happening. If it all started in a comment thread, the answer goes in that thread and on the page. If the episode was born from a video, the page alone will not do — the article on other networks explains why.

The other surfaces get one sentence and a pointer, not a rewrite. "We have published our answer here", with the link. Two different formulations of the same fact immediately produce the accusation of having two stories.

The rule also holds for individual replies during the episode: you answer people by pointing them at the single message, and you handle particular cases privately. It reads as cold and it is what prevents the multiplication described in the previous section.

Finally, the person speaking must not be the person being blamed, for the reason the hub gives: a reply written by somebody defending themselves reads that way, and at this moment everything gets read.

A conditional apology is not one. "We are sorry if some people may have felt…" is read as a manoeuvre by everybody, including people who were not angry with you.

A useful apology has three parts and none of them is emotional. What happened, said plainly. What it cost the people concerned, acknowledged without being minimised. And what changes from now on, with a concrete, checkable modification.

The third part is the only one that genuinely counts. An apology with no change reads as a formula; a change with no apology is often accepted. If you have to choose, choose the change.

The moment is a decision too: you do not apologise before checking, and you do not apologise a week later either. The right window opens once the facts are established and closes quickly, which is another argument for moving fast at the verification stage.

And there is a case where you should not apologise: when you did nothing wrong. Saying so calmly, with the facts, without apologising, is a tenable position — provided you have checked, and provided you do not hold it in a reproachful tone.

Some episodes are not communications subjects and treating them as such makes things worse. They are recognisable by four markers, and in those cases the first decision is to bring in somebody else.

Safety and health: a product that injures, a foodstuff called into question, a dangerous installation. Communication comes after the material measure — withdrawal, shutdown, inspection — and never before. The articles on healthcare and on food and drink describe what those sectors cannot say.

A named accusation against an individual, in either direction. An employee publicly accused, or accusing. Those are facts touching individuals, and a badly calibrated company answer can create liability and damage a real person.

Personal data: unauthorised access, a leak, publication of customer information. The subject is regulated and the response carries legal obligations, including informing the people concerned. It is not a communications decision.

And money: an accusation of fraud, a compromised payment method, a dispute with an amount attached. There too, the public answer waits for the accounting check, and the article on financial services recalls what cannot be written lightly.

What we will refuse first: writing a denial about a fact nobody has verified. It is the commonest request of the first hour, it is understandable, and a denial later reversed costs more than the fact it denied.

We will refuse to draft a public accusation against your carrier, your supplier or a subcontractor. The audience has a relationship with you and not with them, and that sentence reads as an admission that you do not control what you sell.

We will also refuse to run a campaign designed to push a piece of content down the feeds. It does not work, it shows, and the budget spent at that moment is the one that would have fixed the cause.

What we do: the conduct page written on a calm day — the threshold, the names, the numbers, the single place; the holding statement drafted while the internal check runs; the internal instruction given to the teams with its reason; the referral out of communications when the subject requires it; and the conclusion published at the announced deadline.

And what you can do this week, for nothing: write the worst realistic scenario for your business in three lines, then note the name of the person who would publish the first message and the place they would publish it. If either is blank, you have just found the only work that matters before it is needed.