Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Social Media Marketing › Social Media Management

Writing a Social Media Crisis Plan

How to write a social media crisis plan: severity levels, who decides and who speaks, holding statements, pausing the queue, the log, rehearsal and review.
Tutorialintermediate14 min readUpdated

Why the plan has to exist before the crisis

The hours when a plan is needed are the hours when nobody has time to write one. Without it, the same questions get argued in a group chat while the comment thread grows: is this serious, who is allowed to post, has legal seen it, why is the sale post still going out. Each of those questions has a sensible answer that can be agreed calmly in advance, and the plan is where those answers live.

A plan also changes what a crisis is. Most social crises start as something the routine could have handled: a rising number of complaints about the same fault, a scheduled post nobody rechecked, a reply written in the wrong tone. The hub makes the point that most crises are failures of the routine that were visible earlier. A plan that names the signals, the owners and the first actions turns many would-be crises into issues that are dealt with before they travel, and turns the rest into procedures rather than scrambles.

The plan sits beside documents the site already covers. The marketing risk register lists the risks to the plan with a prepared response for each; a social crisis is one of those responses, written out in full. The brand voice and cadence guide sets the tone for bad news from the brand and for a crisis in the wider world. The service recovery guide covers putting things right for the customers affected. The plan connects them; it does not repeat them. When the crisis is in the business rather than on social, the tutorial on leading marketing through a business crisis takes the same pause beyond the social queue to every channel marketing runs.

Before you start

  • The escalation path from community management. The community management guide maps who the community manager calls and when. The plan starts where that path ends.
  • A list of past incidents. Every time the inbox was overwhelmed, a post had to be taken down or the press asked about something on social. These are the raw material for your triggers and scenarios.
  • The people who would be called. Names and deputies from social, customer service, communications or PR, product or operations, legal and leadership, with how to reach them outside working hours.
  • Access to every account and the scheduling tool. Who can pause a queue, edit a profile or post in the brand's name on each platform, written down and checked.
  • The voice guide. The tones for complaints, bad news and wider-world events are set there and used here.
  • An afternoon with the owners in the room. The plan only works if the people named in it agreed to it.

The steps

  1. Define what counts as an issue and what counts as a crisis

    Write two short definitions in words the whole team will recognise. An issue is something the routine can absorb with extra attention: more complaints than usual about one thing, criticism of a post, a pointed question from a journalist. A crisis is when the routine cannot cope: the volume or seriousness outruns the team's capacity, people outside the social team must decide what happens, or the brand needs to say something in its own name. Describe the signals the team can actually see, such as a jump in mentions or messages about one subject, the same complaint arriving from people who do not know each other, a journalist or a public figure joining, any mention of safety, injury or personal data, and a post that is travelling beyond the brand's own followers. Keep the definitions on the first page of the plan, because they are what the person on the inbox reads first.

  2. Set severity levels with triggers from your own history

    A common pattern is three levels: an issue handled by the social team, a serious issue that brings in a named manager, and a crisis that brings in leadership. For each level, write the triggers in terms of your own accounts and history rather than any published threshold, because what counts as unusual volume for a small brand is normal for a large one. Go through the past incidents you listed and ask which level each should have been, then write the trigger that would have put it there. Combine signals rather than relying on one: volume plus seriousness, or seriousness alone when safety, a named person or personal data is involved. Note that a situation can move up or down a level, and say who decides when it does.

    LevelWhat it looks likeWho leadsWho is told
    1. IssueMore messages than usual about one subject, handled by the routine with careCommunity managerSocial lead, at the next check-in
    2. Serious issueThe routine is stretched, the subject is sensitive, or the press or a well-known account has noticedSocial lead, with the relevant managerCommunications, customer service and the duty manager
    3. CrisisThe routine cannot cope, the brand must speak in its own name, or safety, legal or personal data is involvedNamed crisis ownerLeadership, legal and every owner on the RACI
  3. Name who decides, who speaks and who joins

    Build a short RACI matrix for the crisis: who is responsible for each task, who is accountable for the decision, who must be consulted and who is kept informed. The decisions that matter are who declares a level 3, who approves public statements, who is the only voice of the brand once a crisis is declared, and who brings in legal, product, customer service and leadership. Name a deputy for each, including out of hours. Statements follow a fast route through the social media approval workflow, with fewer approvers than routine posts and named deputies so that one person on holiday does not stop everything. Staff who are asked about the situation on their own accounts are directed to the official channel and do not comment themselves; the staff social media policy is the place that rule is written. Journalists go to whoever owns media relations, as the media relations guide describes, and the media enquiries tutorial sets out how that owner logs, triages and answers them and when an enquiry is handed back to this plan.

  4. Write the first-hour actions

    The hub lists the first-hour actions, and the plan turns them into a checklist with an owner against each. Pause scheduled posts on every platform. Acknowledge the situation in public without speculating about causes. Move individual cases to private channels, after acknowledging them in public. Start the log. Tell the people the severity level says must be told. Check that no paid campaign is still running copy that now reads badly, and pause it if so. Add the access details for each action, because the first-hour checklist is useless if the person on duty cannot log in to the tool that pauses the queue. Keep this checklist to a page, printed and saved where the on-call person can reach it without logging in to anything.

  5. Draft holding statements and their approval route

    A holding statement is the first public word: the brand knows, it is taking the matter seriously, here is what people affected should do now, and here is when the brand will say more. Draft one for each scenario the team listed, in the tones the voice guide sets for bad news and for crises in the wider world, and get each one approved in advance by the people who would approve it on the day, so that only the details need filling in.

    ScenarioWhat the holding statement coversWho else approves
    Service outage or delayWhat is affected, where to check for updates, when the next update comesOperations owner
    Product fault or safety concernWhat to do or stop doing now, how to get help, that more will followProduct owner and legal
    Offensive or mistaken postThat the post was wrong, that it has been removed, an apology without excusesSocial lead and communications
    Staff behaviourThat the brand is aware and looking into it, with no comment on individualsHR owner and legal
    A wider-world eventUsually nothing beyond pausing; a message only if the brand has something specific and useful to addCommunications and leadership
    An error made by an AI toolWhat was wrong and the correct information, as the AI mistake guide sets outOwner of the tool and communications

    Corrections to content the brand published follow the guide to updating and correcting content: say what was wrong, fix it in place and keep a note. Mistakes that came from an AI tool also follow the guide to handling an AI mistake. Formal statements to the press belong to the PR owner and the online newsroom. Data breaches, safety incidents, recalls and anything with a regulatory angle go straight to their legal and operational owners, whose rules decide what is said and when; the social plan claims no rule of its own for them and adds nothing beyond an acknowledgement until those owners approve.

  6. Plan how to pause and restart the queue

    Write down, for each platform and each scheduling tool, exactly how scheduled posts are paused, who can do it and how paid campaigns are paused alongside them. Test it: the most common failure in a rehearsal is that the person on duty does not have the access, or the queue lives in a tool only one person can open. Then write the restart rule. The queue does not resume automatically when the situation calms; someone reviews every scheduled post against what has happened, removes or rewrites anything that now reads badly, and the crisis owner agrees the restart. A cheerful product post resuming in the middle of an unresolved thread undoes the work of the holding statement. When the crisis is an account in someone else's hands, the queue may not be reachable at all, and the compromised marketing account response SOP runs the recovery alongside the plan's public side.

  7. Set up the log and the monitoring

    The log is a single shared record of what happened and when: the first signal, each decision and who made it, each public post and reply, each moderation action, each internal message that matters. Use a simple template with a time, an action, an owner and a note, and start it at level 2. It serves three purposes: it keeps everyone working from the same facts during the crisis, it is the evidence for the review afterwards, and it helps answer questions later about what the brand said and did. Monitoring tells the team whether the situation is growing or fading: mentions of the subject, messages in the inbox, comments on the brand's own posts and any press interest. The social listening set-up routes spikes to the crisis owner and gives the team terms to watch. Read the trend by a person, on a sample, rather than trusting an automatic sentiment score.

  8. Agree when and how to close

    A crisis that is never closed leaves the queue paused and the team on edge. Write the closing criteria: the volume of messages about the subject has returned to the team's normal range, the affected customers have a resolution path, no new angle has appeared in the press, and the owners on the RACI agree. Closing means a final public update where one is owed, the queue restarted under the restart rule, the log completed, and the level stepped down so that any remaining messages go back to the routine. Name who announces the close internally, so that everyone who was told about the crisis is also told that it is over.

  9. Rehearse with a scenario

    A plan nobody has used is a guess. Run a tabletop exercise using scenario planning: pick one of the scenarios from the holding statement table, write a short timeline of events, and walk the named people through it, with the facilitator adding new developments as the exercise goes on. Make people do the real things where possible: find the plan, reach the deputy, open the tool that pauses the queue, fill in the holding statement, log the decisions. The academy lesson on handling a public complaint wave for a train ticket app is a ready-made practice case. Write down everything that did not work, fix it, and put the next rehearsal in the calendar before everyone leaves the room.

  10. Review after every real use

    After every level 2 or level 3, hold a short review using the log. Ask what the first signal was and when it was seen, how long each step took, where the plan helped and where it got in the way, and what the routine should change. Use five whys on the cause rather than stopping at the first answer: a late response may come back to a gap in cover, which comes back to a rota nobody owned. Update the plan, the triggers, the holding statements and the access list, and add the incident to the list of past incidents used to set the triggers. Report the review's outcome alongside the usual social report, so that the changes are tracked like any other piece of work.

A ferry operator's plan (illustrative)

Illustrative example — not a real result

A ferry operator writes its first plan

Driftlark Ferries, an invented operator with 3 routes and 2 social platforms, cancelled an evening sailing for a mechanical fault. The cancellation was announced on the website, but the social inbox was not watched after 6pm, and by the next morning there were 140 unanswered comments and messages, a scheduled post promoting summer day trips had gone out at 8pm, and a local reporter had quoted two of the comments.

The marketing manager and the operations director wrote the plan over 2 afternoons. Level 1 covers routine disruption, handled by the inbox. Level 2 starts when a cancellation affects more than one sailing or messages about one subject pass what the inbox usually sees in a whole day, and brings in the duty operations manager. Level 3 starts with any safety concern or press interest, and brings in the director and the company's legal adviser. Five holding statements were drafted and approved, one each for cancellations, delays, a safety concern, a wrong post and a staff complaint.

In the first rehearsal, a 90-minute tabletop exercise based on a fog delay, the out-of-hours duty manager could reach everyone on the list but could not pause the scheduled posts: the scheduling tool had 1 login, held by the marketing manager. Before the summer timetable started, the duty managers were given access to pause the queue, the pause instructions were printed in the duty folder, and a rule was added that the evening's scheduled posts are checked against the day's sailings before 4pm. The next disruption, a cancelled morning sailing, was acknowledged on social 20 minutes after the website notice, the queue was paused, and it stayed at level 1.

Verify it worked

  • Test the triggers against your list of past incidents. Each incident lands at the level you now think it should have had.
  • Ask the person on duty who declares level 3, who approves statements and who speaks. Each role has a named deputy, including out of hours.
  • Open the holding statements. Each scenario in the table has an approved draft needing only details.
  • Run the first-hour checklist from a cold start. The person on duty finds it, logs in and pauses the queue on every platform.
  • Read the rehearsal notes. They show what failed, the fixes made and a log with time, action, owner and note.

If it goes wrong

  • Nobody on duty can pause the queue. Access probably sits with one person or tool. Share access, print the instructions and test again.
  • People argue over which level applies. The triggers were probably copied from another brand or written loosely. Rewrite them from your own incidents and combine signals.
  • Statements go out slowly. The route probably has too many approvers or no deputies. Use the fast route, name deputies and approve holding statements in advance.
  • Scheduled posts resumed mid-thread. The restart probably happened automatically. Check every scheduled post against events and have the crisis owner agree the restart.
  • The matter involves a breach, a safety incident or a recall. It belongs to other owners. Route it to them at once and post nothing beyond an acknowledgement until they approve.

Next steps

Common mistakes

  • Triggers copied from someone else's plan. Thresholds that suit a large brand make a small brand wait too long, and the reverse. Set them from your own history.
  • A plan that names jobs but not people. "Communications approves" fails at night; a name and a deputy do not.
  • No access for the person on duty. The most common rehearsal finding is that the person who must pause the queue cannot.
  • Speculating in the first statement. A cause announced too early and later withdrawn becomes the story.
  • Restarting the queue automatically. Every scheduled post is checked against what has happened before anything resumes.
  • Treating legal and safety matters as social ones. Breaches, recalls and safety incidents go to their owners at once, and their rules decide what is said.
  • Never closing. Without closing criteria the team stays in crisis mode and the routine never resumes.
  • Rehearsing once and filing it. People, access and tools change; the plan needs testing whenever they do.
  • Skipping the review. The cause of the next crisis is usually in the log of the last one.

Frequently asked questions

What is the difference between a social media issue and a crisis?

An issue is something the normal routine can handle with a little more attention: a cluster of complaints about the same thing, a post that drew criticism, a question from a journalist that the press owner can answer. A crisis is when the volume, the seriousness or the attention outruns that routine, so decisions have to be taken by people outside the social team and the brand has to say something in its own name. The plan writes the difference down in terms the team can recognise from its own signals, so that nobody has to argue about it while it is happening.

What should a holding statement on social media say?

That the brand knows, that it is taking the matter seriously, what people affected should do now and when the brand will say more. It does not speculate about causes, assign blame or promise an outcome the business has not agreed. Draft one for each likely scenario in advance, approved by the people who would have to approve it on the day, so that the first public word goes out quickly and only the details need filling in.

Does a social media crisis plan cover legal and regulatory duties?

No. A data breach, a safety incident, a product recall or anything with a regulatory or legal angle has owners elsewhere in the business, and their rules decide what is said, when and to whom. The social plan names those owners, routes the situation to them at once and waits for their approval before saying anything beyond an acknowledgement. Nothing in this tutorial is legal advice.

Terms used on this page

Frameworks

Where this page is used

Also related

Related