Incident Postmortems in Canada: OSFI and PIPEDA Timelines
Run a blameless postmortem that also meets OSFI's 24-hour incident report and the PIPEDA breach record rules, with an ET timeline and a worked example.
By Redbison Dev · · 6 min read

Canada edition. Written for Canadian teams: OSFI's Technology and Cyber Security Incident Reporting advisory (24 hours, updates and a post-incident review) and PIPEDA's real risk of significant harm test and two-year breach records. US edition
If you work at a bank, insurer or trust company regulated by the Office of the Superintendent of Financial Institutions (OSFI), a serious outage starts two clocks at once. OSFI's Technology and Cyber Security Incident Reporting advisory asks for a written report within 24 hours, regular updates and, once the incident is closed, a report on your post-incident review and lessons learned. If personal information was exposed, the Personal Information Protection and Electronic Documentation Act (PIPEDA) adds a breach assessment and a record you must keep for two years. A well-run blameless postmortem is the document that feeds all of these.
This edition works through an outage at a federally regulated financial institution using the Incident Postmortem Generator, with a timeline in Eastern Time and a privacy exposure in the middle of it. It summarizes the rules as OSFI and the Office of the Privacy Commissioner of Canada (OPC) publish them; your compliance and legal teams decide how they apply to you, and provincial privacy laws can add duties of their own.
What OSFI expects, and when
The advisory took effect on August 13, 2021, replacing the 2019 version, and applies to federally regulated financial institutions (FRFIs): banks, foreign bank branches, life, fraternal, property and casualty insurers, and trust and loan companies.
| Stage | What the advisory says |
|---|---|
| Is it reportable? | Any one of a list of characteristics, including: the incident management team or protocols were activated; it was reported to the board or senior management; it was reported to the OPC, another federal department such as the Canadian Centre for Cyber Security, or law enforcement; it is ranked Priority, Severity or Tier 1 or 2 internally; or it breaches internal risk appetite. When in doubt, consult your Lead Supervisor. |
| First report | In writing, to OSFI's Technology Risk Division and your Lead Supervisor, within 24 hours, or sooner if possible, using the OSFI incident reporting form. Unknown details are marked "information not yet available" with best estimates. |
| While it runs | Regular updates (the advisory gives daily as an example) until all details are provided, including short- and long-term remediation plans. |
| After closure | A report to OSFI on the post-incident review and lessons learned. |
| If you don't | Increased supervisory oversight, such as enhanced monitoring, watch-listing or staging. |
That last row of the process is your postmortem. The OSFI criteria also mean that internal severity labels carry regulatory weight: if your runbook calls something Severity 1, it is reportable by definition, so the time the incident was declared Severity 1 belongs at the top of the timeline.
What PIPEDA adds when personal information is involved
The OPC's guidance on breaches of security safeguards sets out three duties that land on the same postmortem:
- Assess real risk of significant harm (RROSH) by weighing the sensitivity of the information and the probability of misuse. The number of people does not decide it: a breach affecting one person must be reported if the assessment finds a real risk.
- Report and notify. Breaches with a real risk go to the OPC with details and, if known, the cause; when it happened; the information involved; the number of people affected; steps taken to reduce harm; how individuals will be notified; and a contact. Individuals must be notified as soon as feasible after you determine there is a real risk.
- Keep a record of every breach for two years, whether or not it met the threshold. At a minimum, the OPC expects the date or estimated date, the circumstances, the nature of the information and whether the OPC and individuals were notified, plus enough detail to show how you reached your RROSH conclusion.
Read against a postmortem template, the breach record is the summary, impact, timeline and root cause sections with one extra paragraph: the RROSH reasoning. Writing it during the review, while the facts are fresh, is far easier than reconstructing it when the OPC asks.
Worked example: a failover that also exposed account summaries
A federally regulated trust company runs its customer app from Toronto, with an on-call team in Vancouver. Toronto and Vancouver are three hours apart. The postmortem uses ET throughout and says so at the top.
In the Incident Postmortem Generator, enter the timeline one event per line:
08:47 ET - Transfer API errors begin after an automated database failover
08:59 ET - Error-rate alert pages the on-call engineer in Vancouver (05:59 PT)
09:10 ET - Incident declared Severity 1 (reportable under the OSFI criteria)
09:21 ET - Support reports customers seeing other customers' account summaries
09:30 ET - Summary cache disabled; exposure stops
10:12 ET - Failback to the primary database; transfers succeed
13:30 ET - Written report sent to OSFI Technology Risk Division and Lead Supervisor
The generator splits each line into its time ("08:47 ET") and its event. Enter 12 minutes to detect and 85 to mitigate; with the permanent fix released at 16:47 ET the same day, enter 480 to resolve. The report shows 12 min, 1 h 25 min and 8 h.
What the timeline proves:
- OSFI. The advisory does not say when the 24 hours start. Counting cautiously from the first alert at 08:59 ET, the written report at 13:30 ET went in after 4 h 31 min (4 h 20 min after the Severity 1 declaration at 09:10 ET), well inside 24 hours either way. Daily updates follow until the post-incident review is filed.
- The privacy exposure. For nine minutes (09:21 to 09:30 ET), a stale cache served account summaries to the wrong customers. Logs show 312 summaries served to someone other than their owner. Account balances and transaction histories are sensitive financial information, and the probability of misuse depends on who saw them; the privacy officer documents the RROSH assessment in the postmortem and, if it finds a real risk, the company reports to the OPC and notifies the 312 customers as soon as feasible. Either way, the breach goes in the two-year breach record.
- The time zone trap. The on-call engineer's notes said "05:59" because they were written in Pacific Time. Left uncorrected, the alert would appear three hours before the errors started, and the time to detect would read as negative. Convert everything to ET with the Time Zone Converter before the review meeting.
The root cause is not "the failover". It is that the summary cache was not invalidated on failover, and no test exercised a failover with the cache warm. Action items follow from that: "Invalidate the summary cache on any database role change, owner: payments platform lead, due in two weeks"; "Add a failover test with a warm cache to the quarterly resilience exercise, owner: SRE lead"; "Alert when one session receives a summary for a different customer ID, owner: security engineering". The generator groups them as prevention, mitigation and detection, which shows at a glance that each layer got one.
Running the review without blame
The blameless approach (asking what let the change cause harm, not who made it) matters more in a regulated firm, not less. OSFI and the OPC both want an accurate account of what happened. People who fear the document will be used against them leave things out, and the gaps tend to be exactly the timestamps a regulator asks about.
- Hold the review within a week while logs and chat history are still at hand, and share the draft a day ahead.
- Have someone other than the incident commander facilitate, and walk the timeline first.
- Leave unknowns blank. The generator writes "Not recorded" for an empty duration; that is better in a report to OSFI than a guess, and the advisory itself allows "information not yet available".
- Track the action items to closure in the same system as other work, and report their status in the post-incident review you send to OSFI.
For long causal chains, the Root Cause Analysis tool works through 5 Whys and fishbone diagrams, and a failure mode analysis helps when the same system keeps producing incidents.
Tools for this job
- Incident Postmortem Generator: timeline, durations, root cause and grouped action items as Markdown you can attach to an OSFI or OPC submission.
- Time Zone Converter: putting ET, PT and UTC log entries on one clock.
- Root Cause Analysis: 5 Whys and fishbone diagrams.
For the general method behind the review, see the blameless postmortem guide.
Frequently asked questions
How quickly must a bank report a cyber incident to OSFI?
OSFI's advisory requires FRFIs to report a technology or cyber security incident to its Technology Risk Division and their Lead Supervisor within 24 hours, or sooner if possible, in writing on the OSFI form. Regular updates follow, and a post-incident review is reported after closure.
Does every outage have to be reported to OSFI?
No, only incidents with one or more of the listed characteristics, such as a Severity 1 or 2 internal rating, activation of the incident management team, or a report to the board. OSFI encourages a precautionary report when you are unsure.
What is a real risk of significant harm under PIPEDA?
It is the test for whether a breach of security safeguards must be reported to the OPC and individuals. You weigh the sensitivity of the information and the probability that it will be misused; the number of people affected does not decide it on its own.
How long do I keep breach records under PIPEDA?
The OPC's guidance says records of all breaches of security safeguards must be kept for two years, including those that did not meet the real risk threshold, with enough detail for the OPC to check your assessment.
Sources and further reading
- Technology and Cyber Security Incident Reporting, Office of the Superintendent of Financial Institutions, effective August 13, 2021. Reportable incident criteria, 24-hour written report, updates, post-incident review.
- What you need to know about mandatory reporting of breaches of security safeguards, Office of the Privacy Commissioner of Canada. RROSH, report contents, notifying individuals, two-year records.
- Privacy breaches at your business, Office of the Privacy Commissioner of Canada. The RROSH self-assessment tool and record-keeping pages.
- Postmortem Culture: Learning from Failure, Google, Site Reliability Engineering, chapter 15.
- Postmortem Practices for Incident Management, Google, The Site Reliability Workbook.
- Postmortem Process, PagerDuty.
Open the Incident Postmortem Generator
More Documentation & Review
Every tool in the Documentation & Review section, free and in your browser.