Crisis Communication System: AI Pricing Overcharge Incident Executive crisis position Recommended posture: Proactive, accountable, specific, and remedy-first. This is a high-severity trust crisis but a manageable reputational crisis if the company acts before the journalist defines the narrative. The central strategic objective should be: > Customers should hear about the problem from the company before—or at the same time as—they hear about it from the media, and every affected customer should already know exactly how and when they will be made whole. Recommended first actions Within the next hour: 1. Freeze and preserve all relevant technical evidence. 2. Confirm the exact population affected and the precise overcharge calculation. 3. Finalize the refund mechanism and verify it can be completed within 72 hours. 4. Brief the CEO, executive team, customer support leadership, and board chair. 5. Respond to the journalist with a holding statement. 6. Send direct notification to all 8,000 affected customers. 7. Publish a controlled public statement before speculation accelerates. 8. Activate a dedicated incident-response page and customer support workflow. Critical principle: Do not announce a 72-hour refund deadline unless the operations team has already validated that it is achievable. --- 1. CRISIS ASSESSMENT AND CLASSIFICATION 1.1 Crisis severity classification framework Use a five-factor scoring model. Dimension Low Moderate High This incident Customer impact Limited inconvenience Noticeable disruption Direct financial loss High Number affected <100 100–5,000 >5,000 High Trust impact Operational issue Confidence concern Core trust threatened High Media potential Limited Industry interest Significant coverage High Legal/regulatory exposure Minimal Possible Significant Moderate—pending review Classification: Severity Level 4 of 5 — High This should be treated as a Tier 1 internal crisis, requiring executive-level command, because: 8,000 customers experienced direct financial harm. The error lasted 72 hours. The incident involves an AI-powered system, increasing public sensitivity. Customers are already discussing the issue publicly. A journalist has initiated contact. The broader customer base of 65,000 merchants will judge the company's reliability based on its response. What prevents this from being Level 5? Currently: The error has been identified. The system has been corrected. The harm is financially measurable and remediable. There is no indication of malicious conduct. No ongoing customer harm is occurring. However, the classification should immediately move to Level 5 if evidence emerges of: A larger affected population. Delayed or failed refunds. A previous warning that was ignored. An ongoing algorithmic issue. Misleading statements to customers. Executive misconduct or concealment. Regulatory investigation. --- 1.2 Crisis type identification This is a hybrid crisis involving four categories: Primary: Product and technology failure A code error caused the platform to perform incorrectly. Secondary: Customer financial harm Customers were directly overcharged. Tertiary: AI and automation trust crisis The involvement of an AI-powered pricing algorithm creates concerns beyond the individual bug: > "Can customers trust the company's AI systems?" Emerging: Reputation and transparency crisis The company's response will determine whether the story becomes: "A company discovered and fixed a software error and proactively refunded customers." or: "An AI company overcharged thousands of customers and customers discovered it before the company came clean." The second narrative is substantially more damaging. --- 1.3 Reputation risk assessment Use this formula: > Reputation Risk = Harm × Visibility × Perceived Responsibility × Response Quality Current assessment Harm: High Direct financial impact. Visibility: Rising Customer forums are already discussing it and media interest has begun. Perceived responsibility: Very high Customers will reasonably hold the company directly responsible. Response quality: Undetermined This is now the primary controllable variable. Recommended reputation strategy Do not spend communication capital arguing: Whether the error was technically understandable. Whether customers should have noticed. Whether only a small percentage of the customer base was affected. Whether the AI was "not really responsible." Instead: > Own the outcome, explain the mechanism accurately, fix the harm, and demonstrate what has changed. --- 1.4 Stakeholder impact mapping Tier 1: Immediate communication priority 8,000 affected customers They need: Clear acknowledgment. Exact explanation. Refund amount/process. Refund timeline. Confirmation that the problem is fixed. A simple support path. Tier 2: Trust preservation priority 65,000 broader customer base They need reassurance that: Their accounts and transactions are functioning correctly. The incident has been contained. The company has investigated the cause. Safeguards have been strengthened. Tier 3: Operational stability Employees They need: One factual version of events. Clear instructions not to speculate. Clear escalation procedures. Confidence that leadership has control. Tier 4: Governance Investor board They need: Facts. Financial exposure. Refund liability. Root cause. Communication strategy. Reputational risk. Recovery milestones. Tier 5: Narrative environment Customer community and media They need a credible, easily accessible source of truth before speculation fills the vacuum. --- 1.5 Media escalation probability Current probability: High The risk is elevated because: A journalist has already contacted the company. The affected number is significant. "AI overcharged customers" is a simple and compelling headline. Forum discussions provide potential sources for journalists. Likely escalation scenarios Scenario A: Controlled coverage Headline: > "E-commerce software company identifies pricing bug and refunds affected customers." Desired outcome. Scenario B: Critical coverage Headline: > "AI pricing system overcharged thousands of customers." Likely if the company appears slow or defensive. Scenario C: Narrative escalation Headline: > "Customers discovered AI overcharging error before company responded." Likely if the journalist publishes before customers receive direct communication. --- 1.6 Golden hour decision framework Every decision in the next hour should pass four tests. Test 1: Is it true? Never communicate unverified root-cause claims. Test 2: Is it sufficient? Does the statement answer what customers immediately need to know? Test 3: Is it empathetic? Does it acknowledge financial harm without hiding behind technical language? Test 4: Does it demonstrate action? Every communication should include: > What happened → What we did → What happens next Golden-hour command sequence 0–15 minutes Establish crisis command group. Confirm facts. Preserve records. Confirm affected population. 15–30 minutes Approve holding statement. Approve journalist response. Confirm refund execution capability. 30–45 minutes Brief board. Brief employees and support teams. Prepare affected-customer communications. 45–60 minutes Contact affected customers. Respond to journalist. Activate monitoring. Publish public holding statement if necessary. --- 2. CRISIS COMMUNICATION STRATEGY 2.1 Core communication principles 1. Lead with the customer impact Do not lead with: > "We identified a technical anomaly." Lead with: > "Some customers were incorrectly charged more than the listed price." 2. Do not wait for perfect information Communicate confirmed facts now and commit to updates. 3. Separate explanation from excuse A code error explains what happened. It does not remove responsibility. 4. Remedy before reputation defense The strongest reputation strategy is making customers whole quickly. 5. Make the next step unmistakably clear Every affected customer should know: Do I need to do anything? How much will I receive? When will I receive it? Ideally, the answer is: > No action is required. We will automatically issue your refund within 72 hours. --- 2.2 Three-message architecture Message 1: Acknowledge and own > We identified a pricing error that caused approximately 8,000 customers to be charged between 15% and 40% above the listed price during a 72-hour period. We are responsible for correcting this. Message 2: Remedy completely > The error has been corrected. Every affected customer will receive an automatic full refund for the amount overcharged, with refunds completed within 72 hours of our announcement. Message 3: Prevent recurrence > We are conducting a full review of how this error reached production and are strengthening testing, monitoring, and automated safeguards before future pricing changes are deployed. --- 2.3 Tone and positioning Recommended tone Direct + human + accountable + technically competent. Avoid Overly legalistic language. Excessive technical explanations. "Unexpected edge case." "Unforeseen circumstances." "We regret any inconvenience." Blaming AI. Blaming customers. Claiming the system is now "100% safe." Positioning statement > A company that identified a serious error, stopped it, informed customers directly, made them whole automatically, and strengthened the system rather than minimizing the failure. --- 2.4 Communication sequence Phase 1: Private notification 1. Board. 2. Employees and customer-facing teams. 3. Affected customers. 4. Journalist. Phase 2: Public transparency 5. Customer community. 6. Public incident statement. 7. Support FAQ. Phase 3: Recovery 8. Refund completion update. 9. Technical safeguards update. 10. Post-incident transparency report. Important adjustment Do not make the journalist the first external party to receive a complete explanation if affected customers have not yet been notified. --- 2.5 Channel priority ranking Priority Channel Purpose 1 Direct email/in-app notification Affected customers 2 Dedicated incident page Single source of truth 3 Journalist response Prevent narrative vacuum 4 Customer forum Correct misinformation 5 Employee internal communication Message discipline 6 Social media Broad awareness 7 CEO statement Escalation if necessary --- 2.6 Spokesperson selection Primary spokesperson CEO or President, if the organization has a credible and prepared leader. Reason: The crisis involves direct customer financial harm and company trust. Technical spokesperson CTO or senior engineering leader, only for technical questions. Customer operations spokesperson Chief Customer Officer or equivalent, for remedy execution. Spokesperson preparation Every spokesperson should be able to deliver the following 30-second answer: > "We identified a code error in our pricing system that resulted in approximately 8,000 customers being charged above the listed price over a 72-hour period. The error has been corrected. We are automatically refunding every affected customer for the full amount overcharged, and we expect those refunds to be completed within 72 hours. We are also conducting a full review of how the error occurred and strengthening our safeguards to prevent a recurrence." --- 3. COMPLETE STATEMENT LIBRARY Initial holding statement — within one hour We have identified and corrected a pricing error that affected a portion of customer transactions over a 72-hour period. Our current review indicates that approximately 8,000 customers were charged above the listed price as a result of a code error in our pricing system. We take responsibility for correcting this. The affected pricing issue has been resolved, and we are now finalizing the process to automatically refund every affected customer for the full amount overcharged. We will provide affected customers with direct information about their refund and timing, and we expect refunds to be completed within 72 hours of our announcement. We are continuing to review the incident and will share additional information as our review is completed. --- First full statement — within 24 hours Over the past 72 hours, a code error in our pricing system caused approximately 8,000 customers to be charged between 15% and 40% above the price displayed at the time of purchase. The error has been identified and corrected. We are sorry this happened. We understand that customers rely on our platform to handle transactions accurately and reliably. In this case, we failed to meet that expectation. The fact that the error was caused by a code issue does not change our responsibility to correct the impact on every customer affected. Every affected customer will automatically receive a full refund for the amount they were overcharged. Customers do not need to submit a claim or take any action to receive their refund. We expect all refunds to be processed within 72 hours of this announcement. We have begun a comprehensive review of how this error reached production, how it remained active for 72 hours, and what safeguards should have detected or prevented it sooner. As part of that review, we are strengthening our testing, monitoring, and pricing-system controls. We are committed to being transparent about this incident and to earning the continued trust of our customers through our actions. We will provide an update when the refund process is complete and will share further information about the safeguards we are implementing as a result of this review. We apologize to every customer affected. --- Media spokesperson Q&A 20 anticipated questions and answer frameworks 1. What happened? Answer: > A code error in our pricing system caused some customers to be charged above the displayed price during a 72-hour period. We identified and corrected the issue and are automatically refunding all affected customers. --- 2. How many customers were affected? > Our current confirmed count is approximately 8,000 customers. Avoid saying "only 8,000." --- 3. How much money was overcharged? > We are calculating the final aggregate amount as part of the refund process. Every affected customer will receive a full refund of the amount they were overcharged. Do not disclose an unverified total. --- 4. Why did the AI cause this? > The issue was caused by a code error in the system. We are reviewing the interaction between the code, deployment process, and system safeguards to understand exactly how it occurred and how it remained undetected. Avoid anthropomorphizing AI. --- 5. Was the AI system making independent pricing decisions? > We are reviewing the specific system behavior involved. What is clear is that the pricing outcome was incorrect and that responsibility for the system and its controls rests with us. --- 6. Why did it take 72 hours to detect? > That is one of the most important questions in our review. We are examining why our monitoring and safeguards did not identify the issue sooner, and we will strengthen those controls based on what we find. --- 7. Did customers discover it before the company did? If confirmed: > Some customers raised concerns that helped bring attention to the issue. We are focused on understanding the complete detection timeline as part of our review. If unconfirmed: > We are verifying the precise detection timeline and will not speculate before that review is complete. --- 8. Why weren't refunds issued immediately? > We first needed to correct the error and verify the affected transactions so that every customer receives the correct refund. That process is now underway, and refunds will be completed within 72 hours. --- 9. Will customers receive compensation beyond refunds? > Our immediate commitment is to fully refund every overcharged amount. We are focused first on ensuring every affected customer is made financially whole. Do not promise additional compensation without approval. --- 10. Can customers trust your platform? > Customers should judge us by both our performance and our response when we fail. In this case, we made a serious error. We corrected it, are refunding every affected customer, and are strengthening the safeguards that should have prevented or detected it. --- 11. Has this happened before? Never answer from memory. > We are reviewing historical system data as part of our investigation. We will communicate confirmed findings rather than speculate. --- 12. Could this happen again? > We cannot responsibly promise that no technology system will ever experience an error. What we can do is identify the failure, correct it, and strengthen the controls designed to prevent and detect similar problems. --- 13. Did executives know before customers? > We are establishing and reviewing the precise timeline. We will communicate verified facts about the incident and our response. --- 14. Are you blaming a software engineer? > No. Our focus is on understanding the system, processes, controls, and decisions that allowed the error to reach customers. Accountability for the system rests with the organization. --- 15. Are employees being disciplined? > We do not discuss individual personnel matters. Our focus is on correcting the customer impact and strengthening the systems and processes involved. --- 16. How much will the incident cost the company? > Our priority is making customers whole. We are still calculating the total financial impact and will provide appropriate information to relevant stakeholders. --- 17. Have regulators been notified? > We are reviewing all applicable obligations and will engage with relevant authorities as required. --- 18. Are customers' other transactions affected? > Based on our current investigation, the issue has been contained and corrected. We are continuing to verify the full scope and will communicate any additional confirmed findings. Only use this if technically verified. --- 19. Was the company trying to increase prices? > No. This was an error, not an intentional pricing strategy. The incorrect charges have been corrected and customers are being refunded. --- 20. Why should customers believe you now? > Because trust is ultimately earned through actions. Our responsibility is to communicate clearly, refund every affected customer, share what we learn, and demonstrate that we have strengthened the system. --- Bridging techniques Use this structure: Acknowledge → Answer → Bridge > "That is an important question. What we can confirm today is [fact]. The most important point for affected customers is [remedy]." Correct → Clarify → Continue > "I would not characterize it that way. What happened was [accurate description]. We are now [action]." Unknown → Commit → Return > "We do not yet have a verified answer to that question. We are investigating it and will provide an update once the facts are confirmed." --- Questions to avoid answering directly Do not speculate about: Individual employee responsibility. Unverified root causes. Potential lawsuits. Insurance coverage. Future regulatory outcomes. Unconfirmed aggregate financial exposure. Exact internal disciplinary actions. Hypothetical future incidents. Avoiding direct answers does not mean evading responsibility. --- 4. STAKEHOLDER COMMUNICATION SYSTEM Employee communication Team, We are addressing an incident involving a code error in our pricing system that caused approximately 8,000 customers to be charged above the listed price during a 72-hour period. The issue has been identified and corrected. Our immediate priority is ensuring that every affected customer receives a full refund for the amount they were overcharged. We expect refunds to be completed within 72 hours of our public announcement. We are communicating proactively because we believe customers should hear directly from us about what happened and what we are doing to correct it. Please do not speculate publicly or discuss unconfirmed information about the incident. If you receive questions from customers, partners, or media, direct them to [designated channel or spokesperson]. Customer-facing teams should use the approved messaging and FAQ provided by the crisis response team. If you encounter a question that is not covered, escalate it rather than creating an individual response. We will continue to update the team as the refund process and incident review progress. Our responsibility now is straightforward: correct the customer impact, communicate honestly, and learn from the failure so that we build a stronger system.Employee FAQ Did we fix the problem? Yes. The identified error has been corrected, subject to ongoing monitoring. Are all customers affected? Approximately 8,000 customers are currently confirmed as affected. Should customers request refunds? No. Refunds should be automatic. What should I say publicly? Do not speculate. Direct people to the official statement. What if I see misinformation? Send it to the crisis monitoring team. Do not independently argue online. --- Customer communication template We are contacting you because a pricing error affected a recent transaction associated with your account. A code error in our pricing system caused your transaction to be charged above the price displayed at the time of purchase. The issue has been corrected. You were overcharged by [AMOUNT]. You do not need to take any action. We will automatically issue a full refund of this amount to your original payment method. We expect your refund to be processed by [DATE/TIME]. Depending on your financial institution, the refund may take additional time to appear after processing. We are sorry for this error. We understand that accurate pricing is fundamental to the trust you place in our platform, and we are conducting a comprehensive review of the incident while strengthening the safeguards designed to prevent similar issues. If you have questions, please contact [DEDICATED SUPPORT CHANNEL]. We apologize for the error and appreciate your patience while we make this right. --- Investor and board briefing Required structure 1. Incident summary 72-hour pricing error. Approximately 8,000 affected customers. 15–40% overcharges. 2. Current status Error identified. Error corrected. Refund execution underway. 3. Financial exposure Estimated refund liability. Processing costs. Potential compensation. Customer churn risk. 4. Reputation exposure Forum discussion active. Journalist inquiry received. Broader customer base monitoring response. 5. Communication strategy Proactive customer notification. Public acknowledgment. Central incident page. Media engagement. 6. Recovery milestones 72-hour refund completion. Week 1 investigation update. Month 1 safeguards update. --- Regulatory communication approach Do not wait to understand legal obligations, but do not send unnecessary communications before legal review. Recommended approach: 1. Immediately involve legal and compliance leadership. 2. Determine applicable consumer-protection, payments, and disclosure obligations. 3. Preserve all relevant records. 4. Ensure public statements do not contradict regulatory submissions. 5. If notification is required, provide: Incident description. Scope. Customer impact. Containment actions. Refund plan. Prevention measures. --- Partner and vendor communication Key partners should hear directly from the company if they may receive customer questions. Core message: > "We identified and corrected a temporary pricing-system error affecting approximately 8,000 customers. All affected customers are receiving automatic refunds. The issue has been contained, and we are strengthening our safeguards." Provide: A partner FAQ. A designated contact. Instructions not to speculate publicly. --- Community communication Enter the customer forum once with an official account. Do not attempt to win every argument. Post: > We are aware of the concerns being discussed here. We have identified and corrected a pricing error that caused some customers to be charged above the listed price during a 72-hour period. Approximately 8,000 customers were affected. We are contacting affected customers directly and automatically refunding the full amount overcharged. We will share updates as the refund process is completed. Then respond selectively to legitimate questions. --- 5. SOCIAL MEDIA CRISIS RESPONSE Monitoring priorities Monitor: Priority 1 Company name. CEO name. Product name. "Overcharged." "Pricing error." "AI pricing." Priority 2 Screenshots of transactions. Viral customer posts. Journalist commentary. Influencer posts. Priority 3 Misinformation. Fraudulent refund claims. Impersonation. --- Response decision tree Respond immediately when: A customer needs refund information. A factual claim is materially incorrect. A high-reach account is spreading misinformation. A journalist asks a legitimate question. Respond selectively when: Customers express anger. Customers criticize leadership. Customers share negative opinions. Do not engage when: Accounts are deliberately provoking conflict. Content is abusive without factual questions. Repeated bad-faith arguments cannot be resolved. --- Social media holding statement We identified and corrected a pricing error that caused some customers to be charged above the listed price during a 72-hour period. Approximately 8,000 customers were affected. We are contacting those customers directly and automatically refunding the full amount they were overcharged. No action is required to receive a refund. We are sorry this happened and are conducting a full review of the incident while strengthening our safeguards. Updates will be shared here: [INCIDENT PAGE] --- Five comment response templates 1. Affected customer > We're sorry this happened. Affected customers are receiving automatic refunds for the full amount overcharged. Please contact us through [SUPPORT CHANNEL] if you have not received information about your transaction. 2. Angry customer > We understand why you're frustrated. Customers were charged incorrectly, and our responsibility is to correct that. The issue has been fixed and affected customers are being refunded automatically. 3. Non-affected customer > Based on our current review, this incident was limited to the identified 72-hour period and affected transactions. We have corrected the issue and are continuing to verify the full scope. 4. Misinformation > We want to clarify one point: [CORRECT FACT]. Approximately 8,000 customers were affected, and every affected customer is receiving a full refund of the amount overcharged. 5. General criticism > We understand the concern. We made an error that affected customers financially. Our focus is correcting the impact, completing refunds, and strengthening the safeguards that should have prevented this. --- Viral content escalation protocol Level 1: Under 10,000 impressions Monitor and respond selectively. Level 2: 10,000–100,000 impressions Crisis communications lead reviews within 30 minutes. Level 3: 100,000–500,000 impressions Executive communications review within 15 minutes. Level 4: Over 500,000 impressions or major media pickup Activate executive crisis call. Determine: Is the information accurate? Does an official correction need to be issued? Does the CEO need to respond? Does the incident page require updating? --- Influencer and media outreach Do not pay influencers to defend the company. Instead: 1. Identify influential commentators already discussing the issue. 2. Offer factual information. 3. Provide the official statement and incident page. 4. Make an informed spokesperson available. 5. Do not pressure anyone to change their opinion. --- 6. REPUTATION RECOVERY ROADMAP Week 1: Immediate stabilization Objectives Complete refunds. Reduce uncertainty. Demonstrate control. Actions Day 1 Customer notification. Public statement. Journalist response. Incident page. Day 2–3 Process refunds. Provide refund progress updates. Within 72 hours Publish: > "Refund processing has been completed for all identified affected customers." Only make this statement when verified. Day 4–7 Complete preliminary investigation. Update FAQ. Address recurring customer concerns. Week 1 success metrics 100% refund completion. Support response time maintained. No unresolved high-severity cases. Declining negative discussion volume. No contradictory public statements. --- Week 2–4: Transparency and accountability Publish a concise incident review covering: 1. What happened. 2. How it happened. 3. Why it was not detected sooner. 4. What was corrected. 5. What safeguards were added. Do not turn this into a technical white paper designed to obscure responsibility. --- Month 2–3: Active recovery Demonstrate tangible improvements. Potential actions: Independent technical review. Expanded automated monitoring. Deployment controls. Pricing anomaly detection. Enhanced customer alerts. Incident simulation exercises. Customer trust campaign Do not launch a broad marketing campaign claiming: > "We're more trusted than ever." Instead, communicate evidence: > "Here is what changed." --- Month 4–6: Narrative rebuilding The company should begin shifting the narrative from: > "The company that overcharged customers." to: > "The company that experienced a serious failure, responded transparently, and materially improved its systems." Support this with: Product reliability updates. Customer success evidence. Independent validation where possible. Leadership discussion of lessons learned. --- Long-term trust restoration milestones 30 days All refunds complete. Root cause identified. Corrective action plan implemented. 90 days Safeguards independently tested. Customer-impact metrics stabilized. Incident lessons incorporated into engineering processes. 180 days Review customer trust indicators. Measure churn. Measure support sentiment. Evaluate media narrative. Conduct full crisis-response after-action review. --- Crisis prevention and preparedness update This incident should permanently result in a stronger system. Establish a Pricing Incident Protocol Every future pricing-system deployment should include: Before deployment Independent code review. Pricing-boundary tests. Automated anomaly scenarios. Rollback preparation. During deployment Limited rollout. Real-time pricing monitoring. Automatic threshold alerts. Human escalation authority. After deployment Transaction reconciliation. Automated variance detection. 24-hour enhanced monitoring. Establish a permanent crisis command team Include: CEO. CTO. Legal. Communications. Customer support. Finance. Product/engineering. Conduct quarterly simulations Test scenarios involving: AI failures. Incorrect pricing. Data breaches. Executive controversies. Platform outages. --- Final strategic recommendation The single biggest reputational risk is not the original code error. It is the possibility that customers conclude: > "The company knew customers were being overcharged and did not tell us immediately." Your communication sequence should therefore be built around one operational commitment: Directly notify affected customers before the story can be defined by a journalist or forum discussion, and ensure the refund mechanism is already active when the notification is sent. If executed with speed, specificity, and complete refunds, this incident can remain a serious operational failure rather than becoming a long-term trust crisis.