How to Handle a Social Media Crisis.

Most brands do not have a social media crisis problem.

They have a permission problem.

When something goes wrong in public, the quality of the response matters less than people assume and the speed of it matters far more. Sprout Social’s Q2 2026 Pulse Survey, conducted with the research firm Glimpse across 2,250 social media users in the US, UK and Australia between 14 and 20 May 2026, found that 84 percent of consumers say the speed of a brand’s response affects how they subsequently perceive it. Only 16 percent said it made no difference.

Speed is not a communications skill. It is an operating one.

What decides whether you answer in an hour or in a day is whether a named person is allowed to post without convening a meeting. So the work is not learning a tone of voice for bad days. It is writing the decision rights, the holding lines and the escalation ladder while nothing is on fire.

What counts as a social media crisis

Not every bad day is a crisis, and getting that boundary right is the difference between a team that escalates properly and a team that cries wolf until nobody listens.

A workable definition: a crisis is an incident where the volume, tone or subject of public conversation about your brand has moved beyond what normal moderation can absorb, and where saying nothing is itself a decision people will judge.

Usually not a crisis:

• One angry customer with a legitimate complaint
• A post that lands awkwardly with a small number of people
• Criticism of a competitor that mentions you in passing
• A single post attracting more negative comments than usual

Usually a crisis:

• A product, safety or service failure affecting many customers at once
• A staff member, founder or partner saying something indefensible
• A claim about your brand spreading faster than you can reply to it
• An organised pile-on rather than scattered individual anger

A compromised account sits in its own category. It looks like a crisis from outside, but the first moves are access control and recovery, which is a different sequence with different owners, and we set that one out in our guide to what to do when a brand account is hacked.

Speed is an operating capability, not a writing one

Every brand believes it would move fast. Then the day arrives and the first ninety minutes go on working out who is allowed to approve a sentence.

That 84 percent figure is not really a finding about consumer psychology. It is a finding about your org chart, because response time is decided almost entirely by how many people have to agree before anything can be published.

If the answer is one person, you can respond in twenty minutes. If it is four people across two time zones plus a legal review, you cannot, however good your writers are.

The approval chain that protects you on a normal Tuesday

Approval chains exist for good reasons. They catch errors, they keep claims accurate, they stop something going out that the legal team has not seen.

On a normal Tuesday that chain costs a few hours and prevents the occasional mistake. It is a sensible trade.

On a bad day the trade inverts. The chain now costs you the only window that matters, and the mistake it is protecting you from is smaller than the mistake of silence.

Most brands never notice the inversion, because they only ever test the chain on normal Tuesdays. The process is not wrong. It is single-mode, and a crisis is the other mode.

Decision rights: name a person, not a department

“Marketing will handle it” is not a decision right. It is a way of feeling covered without anyone actually being accountable at 9pm on a Sunday.

Write down, by name:

• Who can publish a holding response with no further approval
• Who deputises when that person is unreachable, and who deputises for the deputy
• Who can pause the publishing schedule across every account
• Who decides an incident has moved up a tier
• Who is the single point of contact for legal, and what their maximum turnaround is

Give every approver a maximum turnaround as well. An approver with no time limit is a veto, and a veto is what turns an hour into a day.

The test is simple. Pick a random evening, ask who could post right now, and see whether you get one name or a discussion.

Build the escalation ladder while nothing is on fire

An escalation ladder maps severity to who gets contacted and what they are allowed to do without asking. Three tiers covers almost every brand.

1. Tier one: a spike in negative sentiment on a post, or a recurring complaint. The community manager handles it in channel with approved responses, logged for the weekly review.
2. Tier two: a real issue affecting multiple customers, or criticism travelling beyond your own followers. The social lead publishes a pre-agreed holding line within the hour, notifies the marketing lead, and the schedule is paused.
3. Tier three: safety, legal, discrimination, a founder or partner issue, or anything with press attention. The named executive owner is contacted immediately and the response goes out on a single approval, not a committee.

Write it as one page with phone numbers on it. A ladder buried in a slide deck nobody can find at the weekend does not exist.

Write the holding lines before you need them

A holding line does one job. It proves you have seen the thing and are dealing with it, which buys the hours you need to find out what actually happened.

It does not admit fault, speculate on cause or promise an outcome. It acknowledges, it commits to an update by a stated time, and it says where that update will appear.

Draft two or three in advance for the incident types you can reasonably anticipate: a product or service failure, a customer safety issue, a conduct issue involving someone associated with the brand. Get them approved once, while everyone is calm.

Then store them where the on-call person can reach them from a phone. That approval happens now, or it happens at the worst possible moment.

What the first hour actually looks like

The first hour is a sequence, not a debate. One person runs it and everyone else feeds it.

1. Confirm the facts you actually have, and separate them from the ones you are assuming.
2. Assign the tier and name the owner out loud, so nobody is unsure who is running it.
3. Pause anything scheduled to publish.
4. Publish the holding line on the platform where the conversation is happening.
5. Brief whoever answers inbound messages, so the DMs and the public post say the same thing.
6. Set the time of the next update and keep it, even if the update is that you still do not know.

Only one of those six steps is writing. The rest are operational, and each one is a permission somebody has to hold in advance, which is why an incident ladder belongs alongside your everyday social media approval process.

Pause the schedule

Scheduled content is the most common self-inflicted wound in a crisis. A cheerful post landing an hour after a serious complaint reads as contempt, and the screenshot of it outlives the incident.

Know exactly where the pause control is on every platform and in your scheduling tool, and know who can reach it from a phone at short notice.

Then decide the restart rule in advance: what has to be true before normal publishing resumes, and who says so. Teams are usually good at stopping and poor at starting again, which is how a two-day problem becomes a fortnight of silence.

Reply where the conversation is

The traditional instinct is a statement on the newsroom page and a note to journalists. The same Sprout Social research found that 64 percent of consumers think it is important for brands to post on social when a crisis occurs, rather than respond through a press release or a website statement, and that this held across every generation surveyed from Gen Z to baby boomers.

It also found half of consumers are most likely to hear about a brand crisis first on social media, more than double the share who name news articles.

The operational consequence is blunt. If people find out on social and expect the answer on social, your social team cannot be the last department told.

Post the response as a native post on the platform carrying the conversation, not a photograph of a letter. An image of a signed PDF from a name nobody recognises reads as distance, and distance is usually what people are already accusing you of.

Decide the comment policy before the comments arrive

The moment a response goes live, the replies underneath it become the story. Whether you hide, delete, reply or leave alone is a choice people will screenshot, so make it once, in advance, and apply it the same way every time.

A defensible default:

• Leave criticism up, including harsh criticism, and answer the substantive version of it once
• Hide or remove only what breaks a published rule: abuse, threats, illegal content, personal data
• Never remove a comment purely because it is inconvenient
• Reply publicly to the recurring question instead of fifty times individually

Consistency matters more than the exact policy. Selective deletion is what turns one bad day into a second story about a cover-up.

What most brands get wrong

• Treating it as a writing problem. The draft is rarely the bottleneck, the approval is.
• Waiting for complete information. You can acknowledge long before you can explain, and acknowledgement is what the clock is measuring.
• Over-correcting. A long emotional statement about a small problem creates a bigger problem.
• Going quiet between updates, so other people fill the gap for you.
• Letting the most senior person become the writer, which is slow and reads like it was drafted by a committee, because it was.
• Never testing the ladder, so the first live use is also the first time anyone finds out a phone number is out of date.

Rehearse it, or it is not a process

A document nobody has used is a plan. A document a team has run under pressure is a capability.

Twice a year, take an hour. Invent a plausible incident for your category, hand it to the team with no warning, and time the sequence. Nothing real gets published. You are timing how long it takes to reach an approved holding line and a paused schedule.

The same three failures show up nearly every time: nobody was certain who owned it, the approver was unreachable, and the holding lines were stale because the product had changed.

All three are fixable in an afternoon. None of them is fixable on the day.

The recovery window is shorter than you think

Handling it well is a commercial point, not just a moral one. The same research found that when a crisis is resolved properly, 51 percent of consumers would consider buying from that brand again within a few months, and 20 percent within a few days.

That is worth sitting with, because it says the incident itself is rarely what does lasting damage. The handling is.

It also means the last phase deserves as much process as the first. Publish what you found and what changed, once you genuinely know. Close the loop with the people who complained loudest.

Then log the incident honestly: what was said, when, by whom, and what slowed you down. The review is where the ladder gets better, and skipping it means relearning the same lesson at full price.

What good looks like

• One named owner per tier, each with a deputy, both reachable outside working hours
• Holding lines approved in advance and stored somewhere phone-accessible
• A first acknowledgement measured in tens of minutes, not days
• A publishing pause any senior person can trigger without asking permission
• One moderation policy, applied identically across every account
• A stated time for the next update, and updates that arrive when promised
• A written review inside a week, with a named owner for each change

None of that needs a bigger team. It needs the decisions to have been made before they were needed.

How NBK thinks about crisis response

NBK is a social operations partner, so this is the same question we ask about everything else: where does the system slow down, and who is allowed to decide.

A brand that cannot ship a routine post in under two days will not ship a crisis response in under two hours. The speed you have on the worst day is the speed you built on the ordinary ones, which is why we treat this as part of running social like a system rather than as a separate emergency plan filed away somewhere.

The work is unglamorous. Name the owners, cap the approval times, write the ladder, store the lines somewhere reachable, rehearse twice a year.

It is also the only version of this that holds up when it is actually needed.

Next step

If your social process is slowing down good ideas on a normal week, it will not rescue you on a bad one. NBK can help rebuild the workflow behind the content, starting with who is allowed to decide what.