Units
1. Introduction to Artificial Intelligence in Social Media Management: Roles, Boundaries, Authentication, Copyright and Privacy 2. Content Strategy and Content Calendar: Planning with Artificial Intelligence 3. Content Production: Text, Title, Subtitle, Hashtag and Format Adaptation 4. Visual and Video Content: Concept, Scenario, Short Video and Productive Visual 5. Community Management: Comment, Message and Reply Management 6. Comment and Sentiment Analysis: Taking the Pulse of the Community 7. Trend Following and Opportunity Seizing: Newsjacking, Voice and Agenda Management 8. Influencer Collaborations and Campaign Management 9. Analytics and Reporting: Metrics, Insight and Executive Reporting 10. Platform Rules, Originality, Copyright and Brand Security 11. Crisis Communication and Reputation Management on Social Media
Unit 11 / 11

Crisis Communication and Reputation Management on Social Media

Gains:

  • Ability to grade the crisis according to spread, seriousness and responsibility and escalate it to the right level and people
  • Ability to produce fast, empathetic and verified initial response drafts with artificial intelligence and avoid unapproved legal testimony
  • Ability to prepare a guide before the crisis and communicate with one voice, transparent and consistent

How a crisis is managed on social media is the signature of that brand that will be remembered for years. A crisis is an event that threatens a brand's reputation quickly and at scale: a product bug, a bad customer experience going viral, an employee bug, a misunderstanding of a sensitive post, or an external attack (coordinated negative campaign). The character of the crisis on social media is its speed: an outrage can turn into thousands of shares in a few hours. That's why two things matter in crisis communication — speed and the right tone. Artificial intelligence is a powerful assistant here: it summarizes the situation, draws out possible scenarios, drafts responses, sets the tone. But the moment of crisis is the area where AI can "draft" the most, and the decision and the final say definitely belong to the senior management and communications officer. A single wrong sentence can add fuel to the fire rather than extinguishing the crisis.

Recognizing and rating the crisis early

Not every negative comment is a crisis. The first step in crisis management is rating: distinguishing whether the signal is truly a crisis or routine dissatisfaction. Three things increase the likelihood of a signal becoming a crisis: the speed at which it spreads (how many people are talking, how fast), the seriousness of the issue (safety, health, ethics, discrimination), and the responsibility of the brand (is it our fault). An issue that is widespread, serious, and for which the brand is responsible is the highest level and is immediately escalated.

Level

symptom

Response approach

who rules

1 - Routine

Single complaint, low spread

Standard community management

community manager

2 - Ascendant

Increasing negative, theme forming

Watch, prepare, draft

Administrator

3 - Crisis

Rapid spread, serious matter

Official statement, fast and controlled

Senior management + communication

4 - Great crisis

Viral, media, legal dimension

Crisis desk, multi-channel

Crisis team + legal + CEO

Attention: When explaining the situation to AI during a crisis, avoid sharing real names, legal details and information that has not yet been verified. The draft that the AI ​​produces may contain a definitive admission (“we admit our mistake”), a statement of legal liability, or unverified information; Never post them without approval.

Golden rules of crisis communication

A good crisis response is based on several principles. Speed ​​but not rush: respond quickly but not with misinformation; The first response may be "we are aware of the situation, we are examining it". Empathy first: focus on the person affected, not the defense. Transparency: what you know, honestly. Responsibility: if it is your fault, admit it, but confirm the legal liability statements with the law. Consistency: the same message across all channels. One voice: let whoever speaks, let there be no cacophony.

Step by step: crisis flow

  1. Identify and rate. Evaluate the signal for propagation, seriousness and liability; Set the level.
  2. Raise. If level 3+, engage senior management and communications/legal immediately.
  3. Gather information. What actually happened? Collect verified information; Don't speculate.
  4. Prepare the first response. Brief, empathetic, "we are aware and investigating" tone. AI gives draft, human approves.
  5. Manage channels. Monitor comments, DMs; Maintain a single voice and consistent message.
  6. Update. Provide transparent updates as information becomes clear.
  7. Turn it off and learn. Document what happened after the crisis, what you did, and lessons learned.

Four copyable templates

For crisis ratings:

Evaluate the following situation in terms of crisis level (1-Routine, 2-Emerging, 3-Crisis, 4-Major).Criteria: speed of spread, seriousness of the issue (safety/health/ethics), responsibility of the brand.Status (personal data masked): "[situation]". Suggest level, rationale and first 3 steps. Suggest cautious/higher level where unsure.

For the first answer draft:

Write the FIRST draft response to a social media crisis (public, brief).Status: "[neutral summary]". Brand voice: calm, empathetic, taking responsibility but without legal acceptance. It should include: that we are aware of the situation, that we take it seriously, that we are examining it, and that we will inform you as soon as possible. DO NOT post outright admissions of guilt, promises of compensation, or unverified information. Give 3 versions, slightly different tones.

For scenario and question preparation:

Come up with 10 tough questions/criticisms that might come if this crisis escalates and a safe, honest response framework for each. Status: "[summary]". The answers should not be defensive, rather say "we have not been able to confirm yet" about the issue we did not provide information about. Mark the issues that require legal confirmation [ASK THE LAW].

For post-crisis evaluation:

Evaluate this crisis period and make a "lessons learned" note.Timeline and actions taken: [paste]. Output: what went well, what went badly, response time review, 5 suggestions for future improvements, and an early warning checklist.

Weak prompt / Strong prompt

Weak prompt:

There is a crisis, write an apology.

The situation, tone and legal boundaries are unclear; AI produces risky text that is either overly apologetic or defensive; may make an admission of guilt or promise compensation.

Powerful prompt:

Situation (neutral, personal data masked): "Packaging error noticed in a batch of product, several customers complained on social media, root cause not yet confirmed."Brand voice: calm, empathetic, taking charge; NO promise of legal acceptance and compensation.Task: first draft public response (60 words max) — we are aware, we take it seriously, we review it, we will update it. Unverified information and absolute acceptance. Give 2 versions. The final decision is mine.

Secondly, it produces a safe, usable first response because it provides clear status, tone, and legal boundaries.

Tip: Prepare a “crisis guide” before a crisis occurs: level definitions, who notifies whom, who speaks, ready-made initial response skeletons and approval chain. Thinking from scratch in a crisis is the most expensive way; A ready-made frame saves minutes.

three mini cases

Case 1. A food brand used its pre-prepared crisis guide in the event of a product recall; The first empathetic response appeared in 40 minutes, and the verified statement appeared in 3 hours. Thanks to transparent and fast communication, the crisis subsided in 2 days and the brand received comments that it "acted responsibly". AI accelerated the drafts, humans made the decisions.

Case 2. A brand panicked and published the AI-produced sentence "we accept full responsibility and bear the damages" without verification; While the legal process was still unclear, this statement tied the brand and increased the cost. Lesson: every legal statement in a crisis response is confirmed by law.

Case 3. A business remained silent for 6 hours, mistaking a rising tide of complaints for "routine"; This silence created the perception that "they don't care" and the crisis grew. If they had done the rating correctly and given an early "we are aware" message, the wave could have died down. Lesson: silence is also a message; Early and accurate grading is critical.

Common mistakes

  • Wrong rating. Being late thinking the crisis is a routine, or mistaking the routine for a crisis and creating panic.
  • Publishing legal statements without approval. Sharing the crime/compensation admission produced by AI without confirmation.
  • Defensive tone. Self-defense instead of empathy; magnify anger.
  • Providing unverified information. Explaining something that is not yet clear as if it were certain.
  • Polyphonic communication. Different people give different messages; cacophony.
  • Crisis without guidance. Caught off guard and thinking from scratch.

In summary

Crisis communication is the moment when the character of the brand is seen; speed and correct tone are decisive. AI summarizes the situation, generates scenarios and empathetic response drafts, saving valuable minutes — but in the moment of crisis, the decision and final word is left to the human, senior management and communications/legal. Rate the crisis accurately, respond early and empathetically, avoid unverified information and unconfirmed legalese, speak with one voice. The best crisis management starts with a guide prepared before a crisis occurs. With this unit you complete the module: now you can use AI as an assistant at every stage of social media management, always keeping the responsibility on yourself.

Application task

Create a one-page crisis guide for a brand: (1) 4-level rubric, (2) who is in charge at each level, (3) 2 ready-made initial response skeletons, (4) approval chain. Then write a fictional level-3 crisis scenario; Ask AI for ratings, initial answers (2 versions) and 10 difficult questions preparation. Show that you have marked and eliminated legal/strict statements in the drafts.

checklist

  • [ ] I ranked the crisis according to its spread, severity, and responsibility.
  • [ ] I escalated the level 3+ crisis to senior management and legal.
  • [ ] I responded quickly, emphatically, and with verified information.
  • [ ] I did not include any unapproved legal statements or promises of compensation.
  • [ ] I used a single voice and consistent message across all channels.
  • [ ] I provided a transparent update as the information became clearer.
  • [ ] I documented post-crisis lessons and updated the guide.

Module Exam

1. Which principle best summarizes this module's view of artificial intelligence?

  • A) Artificial intelligence is a production and analysis assistant; Final approval and responsibility always belongs to the competent manager ✔
  • B) Artificial intelligence is sufficiently developed, it is the most efficient way to leave the brand account entirely to it.
  • C) Artificial intelligence should only be used in visual production and should not be mixed into text at all.
  • D) Since the artificial intelligence outputs are fluent, they do not need to be verified separately.

Description: The basic principle of the module is that artificial intelligence is a production and analysis assistant; The responsibility and final approval of every post, response and claim published belongs to the competent social media manager.

2. How should the role of artificial intelligence change as the level of risk increases (e.g. crisis initial response, product claim)?

  • A) His role grows, because in critical moments he must decide for speed alone
  • B) Its role becomes smaller; Gives outline only, human approval and final word come first ✔
  • C) Does not change; Used with the same authority in every task
  • D) Completely disables; Artificial intelligence cannot be used at all in high-risk jobs

Description: The rule is clear: as the risk increases, the role of artificial intelligence becomes smaller and human approval grows. In content containing crises and claims, artificial intelligence gives the most drafts and never publishes them alone.

3. What is it called when artificial intelligence fluently produces information that does not actually exist (e.g. a made-up statistic or platform rule) as if it were true?

  • A) Algorithm
  • B) Interaction rate
  • C) Hallucination ✔
  • D) Newsjacking

Explanation: This situation is called hallucination. The model presents non-existent information as if it were true; Therefore, numbers, dates, rules and claims must be verified from the source.

4. What is the right approach when asking AI for help responding to a customer's angry DM?

  • A) Pasting the message into the vehicle as is along with name, phone and order number
  • B) Mask personal information and share only the context of the situation ✔
  • C) Uploading the screenshot with all personal data
  • D) For speed, copy all the customer's information and publish the draft immediately

Explanation: Personal data (name, telephone, order number) is protected under KVKK. The right way is to anonymize the data and share only the context (e.g. delayed shipment, angry tone).

5. What is the main purpose of defining content pillars and value/sales balance when setting up a content calendar?

  • A) Just to produce more sales content
  • B) To prevent artificial intelligence from making up history
  • C) To leave the calendar entirely to artificial intelligence
  • D) To maintain the line of the account and prevent the content from shifting to excessive sales ✔

Description: Content pillars are the main themes around which the account revolves; The value/sales balance (e.g. 4:1) prevents content from shifting towards sales. If the pillar and balance are not defined, the AI ​​output will be sales-heavy and scattered.

6. Why is a statement like "our product the best in the world, unrivaled" risky in an Instagram caption?

  • A) It is an unprovable claim of absolute superiority; Contains trust and advertising regulatory risks ✔
  • B) Penalized by the algorithm for being too long
  • C) There is no risk, strong expressions always increase interaction
  • D) It is risky only if the hashtag is missing

Explanation: Unprovable claims of absolute superiority both undermine consumer confidence and pose risks in terms of advertising legislation. Claims must be measured and evidence-based, and product features must be verified from the product description.

7. What is the highest brand safety risk when using generative imagery (AI image generated from text)?

  • A) Low resolution of the image
  • B) The color palette of the image does not match the brand
  • C) Presenting a real person, event or fake product as real ✔
  • D) The visual should be in illustration style

Explanation: An artificial intelligence visual that portrays a real person, event, or a product of the brand that does not actually exist is misleading and seriously threatens the credibility of the brand.

8. In sentiment analysis, what is the reason why artificial intelligence considers a comment such as "great, it arrived broken again" as positive and what is the correct precaution?

  • A) Confuses irony; ✔ It is necessary to read examples from each class and verify percentages
  • B) The model is always correct; The review is really positive.
  • C) The problem is sample size; the only solution is to add more emojis
  • D) Irony is trivial; percentage can be put directly into management report

Explanation: Sentiment analysis can misread irony, slang, and mixed emotion. Therefore, percentages should not be included in the report without verification; Real samples from each class should be read and labels checked manually.

9. What is the best division of labor when using artificial intelligence to join a current trend?

  • A) Asking artificial intelligence for trends and ideas is the fastest way.
  • B) People discover and verify the trend; AI only generates comments and adaptation ideas ✔
  • C) Using every hashtag told by artificial intelligence without verifying it
  • D) Leaving trend discovery entirely to artificial intelligence and not involving humans at all

Explanation: Since artificial intelligence training data is in the past, it does not know current trends and can make them up. The human discovers and verifies the trend from the platform; Artificial intelligence is used only to interpret that trend and generate unique joining ideas for the brand.

10. Why is it wrong to try to link the brand with a campaign message to the agenda of a natural disaster or tragedy (newsjacking)?

  • A) It is wrong only if it is late in terms of timing
  • B) There is no problem, every agenda is an access opportunity
  • C) It is deemed insensitive and causes serious reputational damage; Tragedy is not a marketing opportunity ✔
  • D) It is risky only if hashtags are not used

Explanation: Tragedy and sensitive agendas are never marketing opportunities; It would be deemed insensitive and cause serious reputational damage. The sensitivity filter always comes before speed and opportunity.

11. What should be considered first when choosing an influencer instead of the number of followers?

  • A) Only with the highest number of followers
  • B) Only based on the visual aesthetics of the content
  • C) Blindly following the list of names suggested by artificial intelligence
  • D) Interaction rate, audience fit, originality and brand safety ✔

Explanation: Number of followers is the least reliable metric. What really needs to be looked at is the interaction rate, audience fit, authenticity (no fake followers/interactions) and brand safety.

12. Why is the "#collaboration" / ad tag mandatory in a paid influencer collaboration?

  • A) Just because it is an aesthetic preference
  • B) Preventing classified advertising is a legal obligation and protects the credibility of the brand ✔
  • C) The tag should actually be avoided because it reduces access
  • D) Only required for mega influencers

Description: In Türkiye and many countries, it is a legal obligation for paid collaborations to be clearly labeled; Sharing without tags is considered classified advertising and causes both legal risk and loss of trust. Transparency is non-negotiable.

13. What is the most important rule regarding numbers in the text produced by artificial intelligence when preparing an executive report?

  • A) Verify each number in the report by comparing it one-to-one with the source dashboard ✔
  • B) If the numbers are in a fluent text, there is no need for verification.
  • C) Allowing artificial intelligence to complete missing metrics
  • D) It is enough to check only the largest number

Description: Artificial intelligence can make up numbers that do not exist in the data table. Each number in the report must be compared verbatim to the source dashboard before submission; A single fake number destroys the credibility of the entire report.

14. What does the concept of "vanity metric" mean and why should we be careful?

  • A) They are the most reliable metrics that have a strong connection with the business goal
  • B) Metrics that measure only conversions and sales
  • C) These are metrics that sound good but have little connection to the business goal; The decision should not be based on these ✔
  • D) These are metrics that are hidden and inaccessible by the platforms.

Explanation: Vanity metrics are numbers that sound nice but have little connection to the business goal (e.g. pure followers, pure likes). The decision should be based on metrics (conversion, saves, clicks) tied to the business goal.

15. What is the most critical point of consideration regarding the initial draft response produced by AI in a social media crisis?

  • A) Since the draft is ready quickly, it can be published directly
  • B) It is sufficient to check only spelling errors
  • C) The draft should be as long and detailed as possible.
  • D) It must not contain unauthorized legal admissions, promises of compensation and unverified information; Final approval must belong to the human being ✔

Description: The moment of crisis is the area where artificial intelligence can give the most drafts. The draft may contain a definitive admission of guilt, a promise of compensation, or unverified information; These should never be published without approval, and legal statements should be confirmed by law.

Previous Finish module