Bad buzz: how to handle a social media crisis

A social media crisis is handled with a protocol written before it happens: grade the severity, know who decides, hold a first-response window, then document what occurred. What turns an incident into a crisis is almost never the initial event. It is the time lost working out who is allowed to reply, and a first reaction written while everyone is still angry. Here is the protocol, in the order it unfolds.

What counts as a social media crisis

A social media crisis is a situation where negative reactions towards a brand escalate beyond the team's normal capacity to respond, and start to threaten reputation, trading or the safety of individuals.

Three situations get confused regularly. Confusing them is expensive, in both directions.

Isolated dissatisfaction is an unhappy customer speaking publicly. That is customer service, not a crisis. Treating it as a crisis gives the message a visibility it would never have had on its own.

Bad buzz is a surge of negative comments around a piece of content or a public statement. It is loud, often short, and turns entirely on the quality of the response.

A crisis commits the company beyond the social channel: product, safety, employment law, a wider social issue. It spills into the press and cannot be settled from a social account.

Grade the severity before reacting

Severity grading is the step that determines everything else: who is alerted, who decides, how fast you respond and on which channels.

It has to run on observable criteria, not on an impression. Mention volume against your normal baseline, the presence of high-reach accounts, media pickup, the nature of the complaint and whether there is legal or human risk are enough to settle it within minutes.

A simple example shows why the step matters. A post picks up forty negative comments in two hours. If the account usually gets five a day, the signal is strong. If it usually gets two hundred, that is an ordinary Tuesday. The same raw number calls for opposite decisions, and only a written baseline lets you see which is which.

The classic mistake is grading by the feelings of whoever first sees the comments. A community manager on the front line almost always overestimates severity, because they receive the entire negative volume personally. A written grid protects the team as much as it protects the brand.

Who decides, and how fast

The decision chain is what wastes the most time in a real situation. It has to be written down by name, with a designated stand-in for every role.

Four roles are enough: whoever detects and raises the alert, whoever grades the severity, whoever approves the message, and whoever speaks. Merging the last two is common and costly, because the person who approves is not always the one who should appear.

On timing, one principle holds in most situations: acknowledge fast, take a position once you know. A message saying the subject is being taken seriously and that a full answer is coming buys several hours without committing the company to unverified facts.

This framework assumes a named on-call contact in the evening and at weekends, which is when escalations most often start. An organisation with nobody reading its comments on a Saturday discovers its crisis on Monday, two days late.

The five reflexes that make a crisis worse

These reflexes have one thing in common: they try to make the problem disappear rather than deal with it.

• Deleting comments or the content. Screenshots already exist. The deletion becomes the story, turning a complaint about a piece of content into a complaint about the brand's honesty.
• Replying in the heat of it. A reply written while annoyed ends up quoted in the coverage. The time it takes for a second person to read it back is never time lost.
• Turning off comments by reflex. It is sometimes necessary, particularly where individuals are being harassed. But it gets decided and announced, it is not a comfort move.
• Replying with copy and paste. Twenty identical replies under twenty different comments make a devastating screenshot and signal that nobody is actually reading.
• Going silent without having decided to. Silence is sometimes the right call, but it is owned internally. Silence by default, because nobody dares to rule, reads as contempt.

Building the first response

The first public response is the one that gets quoted everywhere. It holds four elements and needs no defensive framing.

• What you have understood of the complaint, in the words of the people affected, not the company's.
• What you know at the moment of writing, without pre-empting the conclusions of a check still under way.
• What you are doing, as a concrete, verifiable action rather than an intention.
• When you will come back, with a deadline you will actually hold.

What it leaves out matters just as much: no long justification, no blaming a third party, no promise beyond reach, and no wording that makes the apology conditional on how the other person felt.

What happens internally during a crisis

A crisis consumes people first. It is the least anticipated dimension in most protocols, and the one that leaves the deepest marks once the sequence is over.

The front-line team reads every negative message, often alone and sometimes late in the evening. Planning a handover every few hours, and avoiding leaving one person on moderation for a whole day, belongs in the protocol just as much as the severity grid does.

The second point is internal communication. Employees discover the sequence through social platforms like everyone else, and get asked about it by people around them. A short internal message, sent early, setting out what they can say and what they should redirect to the communications team, prevents a scattering of individual statements that becomes a story of its own.

At Sleeq, this is the element we most often add to the protocols brought to us: they are written for the outside and forget the inside.

Exiting the crisis and running the post-mortem

The exit begins when volume returns to its normal level over several consecutive days. Not when the team is tired of the subject.

The post-mortem happens within two weeks, with the people who lived through it. Four questions are enough: when was the signal first detectable, how long passed before the first response, which decision cost the most time, and what was missing from the protocol.

The output is not a report. It is an updated severity grid, an updated role list and updated response templates. A crisis that does not change the protocol will repeat itself identically.

Preparing before you need it

Preparation costs little and changes everything. A few pages will do: the severity grid, named roles with their stand-ins, numbers reachable outside office hours, three or four first-response templates and the list of sensitive subjects specific to your sector.

That document gets tested. A one-hour exercise, once a year, on a plausible scenario, immediately exposes badly assigned roles. It then lives in the same place as your social media content calendar, because the first decision in a crisis is often to pause scheduled posts.

At Sleeq, a creative social and influence agency based in Paris, on-call coverage is the question that comes up most often when framing a setup. It gets settled at the same time as the in-house versus agency decision, and written into the brief for the pitch. Not every provider covers that scope, as the different types of agencies shows.

FAQ on social media crisis management

Should you delete negative comments?

No, unless they are abusive, defamatory, or target individuals. Deleting a critical but legitimate comment shifts the complaint towards censorship and restarts the sequence, especially since screenshots are already circulating. A moderation policy published in advance, stating what gets removed and why, settles the question without having to rule on it under pressure.

How fast should you respond to bad buzz?

The most useful move is to separate the acknowledgement from the position. The first can go out very quickly and simply says the subject has been seen and taken seriously. The second waits for verified facts. That separation avoids both traps: prolonged silence, and a rushed answer on facts that are still uncertain.

Who should speak for the brand?

It depends on the severity level. On a weak signal, the community manager replies in their professional capacity. On an escalation, the brand speaks, with one message repeated identically everywhere. In a full crisis, a named executive speaks. What matters is that the level of voice is decided in advance, not improvised.

Should you pause scheduled posts?

Yes, from the escalation level onwards. Light content published automatically during a negative sequence creates a contrast that becomes the target of criticism in its own right. The pause has to be immediate and cover every channel, including live paid campaigns, which are often forgotten in the rush.

How do you know a crisis is over?

When mention volume returns to its usual level over several consecutive days and pickup stops. Team fatigue is not an indicator. Resuming posts too early sometimes restarts the sequence, whereas waiting a few more days costs almost nothing in visibility.

Does a small company need a protocol?

Yes, and it costs them less than it costs a large one. Two pages will do: who raises the alert, who decides, who speaks, with the numbers reachable at weekends. The most exposed organisations are often those where one person runs the accounts, because their absence on any given day is enough to leave a situation unanswered.

In short

A crisis is prepared for, not improvised. A four-level severity grid, four named roles, a fast acknowledgement kept separate from the position, and a post-mortem that actually changes the protocol. The costliest reflexes are the ones that try to make the problem disappear rather than deal with it.

To build or stress-test your setup, tell us about your situation or explore our social media offer.