Apologising & Accountability — Unit 18, Upper-intermediate English

**Status update · 09:45 UTC**

We **apologise for** the six-hour disruption to exports and **regret any inconvenience** caused. We **deeply regret** that this release **fell short of expectations** on uptime. The outage was **due to** a misconfigured load balancer after a routine change. The premature deploy was a **lapse in judgment** on our side. Please accept our **sincere apologies** to every team that missed internal deadlines.

Delays in customer comms are **attributable to** an internal handover gap — we **hold ourselves accountable** for that and **accept full responsibility** for slow updates. Data were not exposed while **under our care**, but **affected parties** deserve better **transparent communication**. Hourly updates will continue until we close the incident.

Engineering will **take ownership** of the RCA. **Remedial action** includes failover drills, retraining, and **priority remediation** on data integrity checks. **Rest assured** this is treated as **top priority**. **Going forward**, production changes require dual approval and staged rollouts. No deploy ships on Fridays without executive sign-off.

We will **make amends** with service credits per contract, offered **without prejudice** to other remedies. We will **make right** any customers affected by duplicate charges. **Lessons learned** will feed into change control and quarterly audits. We will update you **at our earliest opportunity** with the final timeline. Thank you for your patience while we complete verification. Status updates will remain hourly until all **affected parties** confirm service is stable.

**Final update · 16:20 UTC · incident closed**

Exports have been stable for eleven hours and the incident is now closed. This final note corrects two statements in our earlier updates, because we would rather be accurate late than consistent and wrong.

**Correction one.** Our 09:45 update stated that the outage lasted six hours. The correct figure is seven hours and forty minutes, measured from the first failed export rather than from the point at which we noticed. We are stating the longer figure because it is the one customers experienced.

**Correction two.** We said data were not exposed. That remains true and we have now verified it rather than inferred it. At 09:45 we had not completed that verification, and we should have written "we have no indication of exposure and are verifying" instead. The statement was accurate; the certainty behind it was not yet earned.

**On duplicate charges.** Fourteen accounts were charged twice. All fourteen have been refunded without needing to contact us, and each will receive a direct message today. If you believe you were affected and have not heard from us by tomorrow, please write to us — we would rather investigate a mistaken report than miss a real one.

**What we are not saying.** We are not describing this as an unforeseeable event. The change that caused it bypassed staged rollout because a manager approved an exception under time pressure, and that manager was following a process that permits exactly this. The individual decision was reasonable; the process that made it possible was ours to fix and we had known about it since February.

**Service credits** are being applied automatically under contract and require no action from you. They are offered **without prejudice** and do not limit any other remedy available to you.

**Lessons learned** will be published in full, including the February note, on 30 June.