Business Continuity Plan
beta · Family standing-standards · Phase undefined · Sizes lean, full · ~6,700 tokens
How an organization keeps its critical business processes going through a significant disruption and recovers them afterward: the team and activation authority, each critical process with its recovery objectives, the manual workaround that covers it, and how the plan itself is tested and kept current. The full variant adds a Disaster Recovery section for recovering the systems those processes depend on, built where no separate disaster recovery plan already exists.
The short card. Why the document is shaped this way, and the argument behind every rule here, is in
business-continuity-plan_companion.md. A fully worked instance is
business-continuity-plan_example.md.
When to use
Section titled “When to use”- A named set of critical business processes needs to keep running through a significant disruption, and recover afterward, and you want that response agreed and rehearsed before the disruption happens, not argued out while it is already underway.
- More than one person might have to activate or run this plan, so the activation authority, a named successor, and the team’s own roles need to be written down rather than carried in one person’s head.
- A Business Impact Analysis already exists, or is being built alongside this plan, and you need a place to carry its Recovery Time Objectives and Recovery Point Objectives into concrete continuity strategies, and, in the full variant, a recovery strategy for each critical system.
- You want the plan rehearsed on a schedule, with an observer independent of the team being exercised, and every finding turned into an owned, dated fix rather than a note nobody revisits.
- You want a document that states the specific events that would make it wrong, with a named owner, instead of a calendar reminder nobody actually reads before renewing it.
- Your organization is regulated, or contracts with suppliers who are, and the plan’s own update and disclosure triggers need to be stated rather than assumed.
When NOT to use
Section titled “When NOT to use”- You need the procedure for one known alert or operational situation, not the standing plan for a significant business disruption. That is a runbook: one operational procedure for a known, narrower situation. This plan’s Disaster Recovery section can call out to a runbook; it is not a bigger runbook itself.
- You are checking a service’s standing fitness to be handed off and run day to day. That is a production readiness review, which starts at a routine handoff. This plan starts once a disruption is declared or a drill begins, not at a routine handoff.
- You are writing up what happened after one disruption that has already resolved. That is an incident postmortem, or the dated record of one activation of this plan. A plan that only describes one real event after it happened is a narrower artifact than the standing plan this bundle templates.
- You are tracking potential future events on an ongoing basis, before any of them has actually happened. That is a risk register. This plan is the designed response once a tracked, or untracked, risk actually crystallizes into a disruption.
- Your “disaster recovery” need is really a technical high-availability design. High availability and disaster recovery are related but not the same thing, and a technical high-availability design, by itself, is not this plan’s Disaster Recovery section.
- You want a standard every increment of work is judged against, on a cadence. That is a definition of done. Both are standing-standards siblings, but a definition of done governs whether an increment of work is finished, and this plan governs whether the organization can keep operating through a disruption.
Pick a variant
Section titled “Pick a variant”Lean (eight sections) is Purpose and Scope, Business Continuity Team and Roles, Business Impact Analysis Summary, Continuity Strategies and Manual Workarounds, Activation and Incident Management, Training, Testing and Exercising, Plan Administration, and Review Trigger. Use it when a separate, standalone disaster recovery plan already exists: Purpose and Scope points to it by name and location, the shape FEMA’s own template allows, rather than this plan duplicating it.
Full (nine sections) adds Disaster Recovery immediately before Review Trigger: the recovery strategy for each critical system, backup and offsite storage including at least one offline copy, recovery procedures written so another trained person could follow them, the recovery order across dependent systems, restore verification, and the recovery test. Use it once no separate disaster recovery plan exists and this plan needs to carry that detail itself.
No source read for this bundle states a sizing signal more specific than that one choice, whether a separate disaster recovery plan already exists. Every lean heading appears in full unchanged, in the same order, so growing from lean to full is additive. You never reorder or rename a section you already filled in.
Quality rubric (self-grade)
Section titled “Quality rubric (self-grade)”Score each 0, 1 or 2. Full below 12 out of 18 ships a plan that reads complete on the page but fails exactly when a disruption actually happens: an authority nobody can reach, a workaround nobody has described, or a recovery nobody has tested. Lean is scored on 8 of these 9 rows only (row 8, Disaster Recovery, applies only to full; see the scope table below), and below 11 of 16 the same failure applies to the smaller shape.
| # | Criterion | 0 | 1 | 2 |
|---|---|---|---|---|
| 1 | Scope is bounded | No disruption is defined, or the plan reads as though it covers everything with no stated boundary | A disruption is defined, but a reader cannot tell whether disaster recovery detail lives in this document or elsewhere | A significant disruption is defined in plain language for this specific business, and the document says, by name, whether disaster recovery detail lives in this plan’s own section or in a separate, named document |
| 2 | Authority named and succeeded | No one is named with authority to activate this plan | An authority is named, but with no successor, or the successor has no stated title | One named job title has authority to activate, a named successor is stated, and the document says what happens if normal channels of direction are themselves disrupted |
| 3 | Impacts and dependencies concrete | Recovery Time Objective or Recovery Point Objective rows read “TBD” or carry no unit of time | Some rows state concrete time units, but dependencies are left as “various” or blank | Every critical process states its Recovery Time Objective and Recovery Point Objective in concrete time units, and names the specific systems and suppliers it depends on |
| 4 | Workarounds describe actual steps | A workaround is asserted to exist, with no description of what anyone actually does | Some workarounds describe steps; others stop at asserting one exists, or no recovery level is explained | Every workaround states what staff actually do, step by step, and the recovery level chosen and why, weighed against its cost |
| 5 | Activation sequence is checkable | No stated activation criteria, no notification order, or no stand-down condition | Criteria and a notification list exist, but the order is unclear, or no method is stated for a channel that might be down | Specific activation criteria are stated in advance, notifications go out in a stated order by a method that still works if the usual channel is down, and a stand-down condition is named |
| 6 | Findings owned and observed | No exercise is recorded, or the only observer is the team being exercised | An exercise is recorded, but at least one finding carries no owner, no date, or the observer is not independent of the team | An exercise type and date are recorded, observed by someone independent of the team being exercised, and every finding carries a named owner and a due date |
| 7 | Plan reachable offline | No location is stated for a copy of the plan, or the only copy sits on the network the plan exists to recover | A location is stated, but it is not clearly outside the systems this plan protects, or no master-copy holder is named | A master-copy holder is named, and a reachable, offline copy is kept somewhere that does not depend on the systems this plan protects |
| 8 | Recovery actually tested (full) | A system is listed with “we have backups” and nothing else | A recovery strategy is named for some systems, but no dependency order is stated, or no restore test has actually been run | Every critical system names a recovery strategy, its position in the recovery order relative to what it depends on, and the result of an actual restore test, not only a scheduled one |
| 9 | Trigger names events | Only a calendar date is named, or nothing at all | An event is named, but with no owner, or an annual exercise requirement stands in as the only trigger | Specific conditions are named (a change in people, suppliers, systems, facilities, the business, or the Business Impact Analysis; an activation; an exercise or test finding), each paired with a named owner, with any required annual exercise stated as a floor under those conditions rather than as the trigger itself |
Which rows apply to what.
| Document | Rows | Maximum | Score against |
|---|---|---|---|
| full | all 9 | 18 | 12 |
| lean | 1-7 and 9 | 16 | 11 |
Row 8 is scored only against full. Lean ships no Disaster Recovery section of its own, carrying a pointer to a separate plan instead, so grading it on that row would penalize the choice of variant rather than the quality of the document.
The test behind every cell above: could someone satisfy it without improving the document? A row that counted workaround rows, vendor rows, or exercise findings would reward padding. Every cell instead asks whether a specific piece of evidence exists, and whether a second person, not the author, could check it without asking who wrote the document.
Named anti-patterns (the usual wrecks)
Section titled “Named anti-patterns (the usual wrecks)”- Neglect, because disasters are rare. SANS’s disaster recovery policy template says it plainly: management often ignores the planning process because disasters happen so rarely. The same rarity is why nobody notices the plan going stale until the day it is needed.
- Disaster recovery bolted on late. A system grows complex and interdependent first, and only once it is already load-bearing does anyone ask how it would recover, by which point the honest answer is “with difficulty.”
- Backups nobody has restored. Having backups is not the same claim as having a recovery. A backup schedule with no restore test behind it is a belief, not a capability, and the gap stays invisible until the restore is the only option left.
- The rarely used path breaks first. The parts of a system that only run during a disaster get no routine exercise, so, as Google engineer Tanya Reilly says in an interview on disaster recovery plans, they need the most deliberate testing. Left untested, they fail at the moment they are needed.
- A test that does not exercise the people who will recover. Watching someone else’s staff run a recovery from backup tapes, or testing only the technical process, confirms nothing about whether your own team could actually execute it under pressure.
- Dependency deadlock during recovery. Two systems that each wait on the other to start can strand a recovery indefinitely if the order of recovery was never written down and checked against what each system actually depends on.
- Stale contacts and documentation. The people, numbers, and procedures a plan names need more frequent review than the rest of its content, and a recovery exercise is one way to find out that a contact list or a procedure has quietly gone out of date.
- The plan itself unreachable during the outage. Keeping the only copy of this plan on the network, or in the facility, it exists to recover means it is unreachable at exactly the moment someone needs to read it. This is a stated precaution in the sources this bundle read, not a measured failure, and it is named here on that honest footing.
Pairing with your process
Section titled “Pairing with your process”This bundle ships in the standing-standards family alongside definition-of-done, definition-of-ready,
launch-coordination-checklist, production-readiness-review and runbook, as a tool: a plan a team
executes under time pressure once a disruption is declared or a drill begins, not a standard the team is
judged against on a cadence. A runbook is the procedure for a known, narrower operational situation;
this plan is the procedure for a significant disruption to the business itself, and its Activation and
Incident Management section is the seam where an organization’s own incident response runbook picks up. A
production readiness review checks a service’s standing fitness to be handed off and run day to day; this
plan does not duplicate that check, and a service’s own backup and disaster recovery responsibility can sit
entirely inside this plan instead, named rather than re-derived. Update this file when the plan itself needs
to change, not every time it is activated or drilled.
The artifacts
Section titled “The artifacts”business-continuity-plan_template-lean.md · ~6,700 tokens
---title: "{{plan_title}}"business_unit: "{{business_unit}}"owner: "{{owner}}"status: "{{status}}"last_updated: "{{date}}"doc_type: business-continuity-plansize: leansource_template: business-continuity-plansource_template_version: 0.1.0---
<!--LEAN BUSINESS CONTINUITY PLAN. The load-bearing core: who this plan covers, the team and the activationauthority, what each critical process needs to survive a disruption, how it keeps running without itsnormal systems, how an activation unfolds, how the plan is tested, the records that keep it usable, andwhat would make the plan itself wrong. Eight sections, because disaster recovery is documented in aseparate plan and Purpose and Scope points to it by name and location. When this plan must carry disasterrecovery itself, add the Disaster Recovery section from business-continuity-plan_template-full.md. ADDthe section, never rename or reorder the ones below, because the full variant is a strict superset of thisone.
THIS PLAN KEEPS BUSINESS PROCESSES GOING; A DISASTER RECOVERY PLAN RECOVERS THE SYSTEMS THOSE PROCESSESUSE. NIST defines the plan as "The documentation of a predetermined set of instructions or procedures thatdescribe how an organization's mission/business processes will be sustained during and after asignificant disruption," and FEMA's Ready Business program ships it as a fill-in template. Disasterrecovery is narrower: NIST scopes it to "an alternate facility." This lean variant points to a separatedisaster recovery plan, the shape FEMA's template allows when it lists "Information Technology DisasterRecovery Plan (if not included in the business continuity plan)." See business-continuity-plan_companion.mdsection 1 and section 6, item 1.
THIS PLAN IS A STANDING INSTRUMENT, NOT A RECORD OF ONE DISRUPTION. A team executes it under timepressure once a disruption is declared or a drill begins. Update this file when the plan itself needs tochange, not every time it is activated. See business-continuity-plan_companion.md section 6, item 3.
WHAT THIS PLAN IS NOTIt is NOT a runbook, which is one operational procedure for a known, narrower situation; this plan can callout to one. It is NOT a production readiness review, which checks a service's fitness for a routinehandoff. It is NOT an incident postmortem, written after one event has resolved. It is NOT a risk register,which tracks events before they happen. And a high-availability design is not disaster recovery by itself.See business-continuity-plan_companion.md section 8.
HOW TO FILL THIS IN1. Read the comment under each heading: WHAT it wants, WHY it matters (with a pointer into business-continuity-plan_companion.md), guiding questions to ASK, a GOOD and a WEAK example, and the TRAP to avoid. For tables, PRIORITY explains the ordering rule and ROW HINT says what a good row contains.2. Replace each {{placeholder}} with your content. Fill Business Continuity Team and Roles and Business Impact Analysis Summary first; the rest depends on knowing who runs this plan and which processes it protects.3. If a section, or one row inside a table, does not apply, write "N/A" and the reason beside it rather than leaving it blank or deleting the section. An unexplained "N/A" and a skipped item look identical to a later reader.4. Before you ship it: self-grade against business-continuity-plan_guide.md, then DELETE every HTML comment. They are guidance, not content.-->
# {{plan_title}}
## Purpose and Scope
<!-- WHAT The business or unit this plan covers, what counts as a significant disruption in this plan's own scope, and where a separate disaster recovery plan lives if this plan does not carry one. WHY NIST defines the type this plan fills: "The documentation of a predetermined set of instructions or procedures that describe how an organization's mission/business processes will be sustained during and after a significant disruption." FEMA's Ready Business template lists "Information Technology Disaster Recovery Plan (if not included in the business continuity plan)" among the related plans a complete program keeps, which is the source for the pointer below: where this plan does not carry its own Disaster Recovery section, this is where a reader is told a separate plan exists and where it is kept. Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Purpose and Scope) and section 6, item 1. ASK What business or unit does this plan cover? What counts as a "significant disruption" here, in plain language? Does a separate disaster recovery plan already exist as its own document? If so, name it and say where it is kept. GOOD "This plan covers Meridian Freight's claims-processing division. A significant disruption is any event, a facility loss, an extended system outage, a regional disaster, that prevents claims processing for more than four hours. A separate IT Disaster Recovery Plan is maintained by the Infrastructure team and is not duplicated here; see the Infrastructure team's internal wiki for the current copy." WEAK "This is our business continuity plan." (no scope, no definition of a disruption, and no note on where disaster recovery lives) TRAP Treating "disaster recovery plan" and "business continuity plan" as two names for one document. NIST and FEMA both treat disaster recovery as a narrower plan that may sit inside or beneath this one; neither source treats the two as equivalent. -->
{{continuity_scope_statement}}
**Separate disaster recovery plan, if one exists outside this document:** {{disaster_recovery_plan_pointer}}
## Business Continuity Team and Roles
<!-- WHAT The team roster by job title, the person with authority to activate this plan, a named successor, and any delegation that takes effect when normal channels of direction are disrupted. WHY FEMA's template carries its own "Business Continuity Organization" section naming the team. California's instructions ask that "job titles (rather than the names of individuals) should be used to assign responsibility as it lessens maintenance on procedures as personnel changes." NIST requires a single activation authority: "Only one individual should have this authority, and a successor should be clearly identified to assume that responsibility if necessary." FEMA's continuity guidance adds the delegation mechanism: "predetermined delegations of authority will take effect when normal channels of direction are disrupted and will terminate when these channels are reestablished," and asks that delegates also be named "by position title and not by name." Who actually writes, activates, and executes a plan like this is not settled across the sources this bundle read; name your own roles explicitly rather than assuming one pattern applies here. Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Business Continuity Team and Roles) and section 6, item 4. ASK What roles does this plan depend on, by job title? Who currently holds each one? Who has authority to activate this plan, and who is that person's named successor? Does a delegation clause need to state what happens if neither is reachable? PRIORITY List the activation authority and successor first, so a reader under time pressure finds the one answer they need before anything else on this page. ROW HINT A good row names a job title, not only a person, and states what that role actually does when this plan runs. A weak row names a person with no title and no stated backup. GOOD | Business Continuity Team Lead | Priya Shah, Director of Operations | Has authority to activate this plan; successor is the Deputy Director of Operations | WEAK | Team Lead | Priya | (no title, no successor, nothing to fall back on if Priya is unreachable) TRAP Naming individuals instead of job titles. California's own instructions give the reason directly: it "lessens maintenance on procedures as personnel changes." A plan that names only people goes stale the day one of them leaves. -->
| Role | Job Title and Current Holder | Responsibility ||---|---|---|| {{team_role}} | {{team_role_holder}} | {{team_role_responsibility}} |
**Authority to activate this plan:** {{activation_authority_title_and_holder}}**Named successor if the activation authority is unreachable:** {{activation_successor_title_and_holder}}**Delegation of authority, if normal channels of direction are disrupted:** {{delegation_of_authority_statement}}
## Business Impact Analysis Summary
<!-- WHAT Each critical business process, its Recovery Time Objective and Recovery Point Objective, and the systems and suppliers it depends on, carried over from a separate Business Impact Analysis. WHY NIST frames the analysis as upstream and prior: "The BIA is the first source for determining resiliency and contingency planning strategies," and its results "should be appropriately incorporated into" the plans that follow. FEMA's own template field reads "Insert results of Business Impact Analysis," and California requires that the recovery plan "must be derived from" the BIA. The two objectives are defined consistently across the sources this bundle read: a Recovery Time Objective is "the maximum acceptable delay between the interruption of service and restoration of service," and a Recovery Point Objective is "the maximum acceptable time after the last data recovery point." A supplier contract states the same obligation this table carries for suppliers: "Supplier shall ensure all Subcontractors have Business Continuity Programs and Disaster Recovery Plans sufficient to meet delivery of the Services during a Disruptive Event." Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Business Impact Analysis Summary). ASK For each critical process: what is its Recovery Time Objective and Recovery Point Objective, in plain units of time? What systems does it depend on? What suppliers does it depend on? PRIORITY List the process with the shortest Recovery Time Objective first, so the most time-critical process is visible without reading the whole table. ROW HINT A good row states Recovery Time Objective and Recovery Point Objective in concrete time units and names specific systems and suppliers, not "various." A weak row writes "TBD" or leaves the dependency columns blank. GOOD | Claims intake | 4 hours | 1 hour | Claims database; the regional scanning vendor | WEAK | Claims intake | Fast | N/A | (no units, and the dependency columns are not actually filled in) TRAP Deriving Recovery Time Objectives and Recovery Point Objectives from scratch in this table. This table carries a separate Business Impact Analysis's results; it is not the analysis itself. If you are deriving these figures for the first time, that work happens in a Business Impact Analysis, not here. -->
| Critical Process | Recovery Time Objective | Recovery Point Objective | Systems and Suppliers Depended On ||---|---|---|---|| {{critical_process}} | {{process_rto}} | {{process_rpo}} | {{process_dependencies}} |
## Continuity Strategies and Manual Workarounds
<!-- WHAT How each critical process keeps running without the systems it normally depends on, and at what recovery level, chosen against cost. WHY FEMA's template names this as its own section, "Manual Workarounds." The cost tradeoff that governs how far to invest in a workaround is stated most explicitly by AWS's own recovery strategies, ordered "in increasing order of cost and complexity, and decreasing order of RTO and RPO," with an explicit warning: "You should avoid implementing a strategy that is more stringent than it needs to be, as this incurs unnecessary costs." Google states the same tradeoff for the process as a whole: "Typically, the smaller your RTO and RPO values are... the more your application will cost to run." A high-availability technical design, by itself, is not a continuity strategy for the business process it supports: "HA doesn't entirely overlap with DR, but it's often necessary to take HA into account when you're thinking about RTO and RPO values." Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Continuity Strategies and Manual Workarounds). ASK For each critical process, what do staff actually do by hand when the normal system is unavailable? What recovery level was chosen, and why, given what it costs against how fast the process actually needs to recover? PRIORITY List the process with the shortest Recovery Time Objective, from Business Impact Analysis Summary above, first, since it is first in line for investment. ROW HINT A good row states what staff actually do, step by step, not that a workaround "exists." A weak row asserts a workaround with no description of the steps. GOOD | Claims intake | Staff log claims to a shared paper form and a backup laptop with offline storage, then enter them into the database once it is back | Manual, low cost, matches the process's own four-hour Recovery Time Objective | WEAK | Claims intake | Manual workaround exists | (no description of what staff actually do, and no stated reason for the recovery level chosen) TRAP Choosing a workaround faster than the process actually needs. AWS's own warning applies here even though it was written about technology recovery: a workaround that beats its process's own Recovery Time Objective is also one that costs more than it needs to. -->
| Critical Process | Manual Workaround | Recovery Level Chosen, and Why ||---|---|---|| {{workaround_process}} | {{workaround_description}} | {{workaround_recovery_level_rationale}} |
## Activation and Incident Management
<!-- WHAT Stated activation criteria, who declares an activation, who is told and in what order, and the condition for standing the plan down. WHY NIST requires that activation criteria be stated in advance and tied to a designated authority: "The designated authority... has the authority to activate the contingency plan," triggered when "one or more of the activation criteria for that system are met." California asks the plan to identify "individual(s) responsible for making the TRP activation decision." HMPPS separates who activates from who writes the plan: "A member of the nominated Business Continuity team will normally activate and stand down the plans." Where regulation requires it, notice extends past an organization's own staff: FINRA binds its member firms to "disclose to its customers how its business continuity plan addresses the possibility of a future significant business disruption." Deactivation is its own named step, not an afterthought: NIST's reconstitution phase "consists of two major activities: validating successful reconstitution and deactivation of the plan." Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Activation and Incident Management). ASK What specific conditions trigger an activation? Who declares one? Who is notified, in what order, and by what method, including anyone outside the organization you owe notice to? What condition stands the plan down? PRIORITY List notifications in the actual order they go out, so the first call someone makes during a real disruption is the first row on this page. ROW HINT A good row names a specific role or entity and a method that still works if the usual channel is down. A weak row says "notify everyone" with no order and no method. GOOD | 1 | Business Continuity Team Lead | Paged directly; confirms the activation decision | WEAK | | Everyone | (no order, no method, and no way to tell who actually gets called first) TRAP Treating a regulatory or customer notification duty as universal. FINRA's duty binds its own member firms specifically; fold a notification line like it in only where your own organization actually owes that notice. -->
**Activation criteria:** {{activation_criteria}}**Who declares an activation:** {{activation_decision_maker}}
| Order | Who Is Notified | Method ||---|---|---|| {{notification_order}} | {{notification_recipient}} | {{notification_method}} |
**Condition for standing this plan down:** {{stand_down_condition}}
## Training, Testing and Exercising
<!-- WHAT The exercise schedule and type, what each one validates, who observes it, and each finding as an owned, dated corrective action. WHY The FFIEC names four test types, from lightest to heaviest: "Orientation/Walk-through," "Tabletop/Mini-drill," "Functional Testing," and "Full-scale Testing," and names a specific insufficient test: "Unmanned recovery testing, where back-up tapes are sent to the recovery site to be run by service provider employees, is not a sufficient test." The same source asks that "the audit department or other independent party should directly observe tests." A test is judged against named criteria, not a bare pass or fail: Google names three, "data integrity, recovery time objective (RTO), and recovery point objective (RPO)." Microsoft adds that a test must exercise the people, not only the systems: "Include all of the human processes in your drills, and don't just focus on the technical processes." Findings become tracked, owned work: FEMA's continuity guidance names a standing "Corrective Action Program (CAP)," where "Deficiencies, actions to correct them, and a timeline for remedy are documented," and HMPPS requires that "Debrief sessions must be held following an exercise and lessons learnt log completed." A practitioner argues directly that these activities carry no pass or fail judgement at all, "these activities are an exercise, not a test"; this bundle keeps both words in this section's own title rather than choosing a side. Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Training, Testing and Exercising) and section 6, item 6. ASK What type of exercise is scheduled? What is it meant to validate? Who observes it, independently of the team being exercised? For every finding, who owns the fix, and by what date? PRIORITY List the most recent exercise first, so a reader can see immediately whether this plan has actually been tested recently. ROW HINT A good row's finding carries a named owner and a due date, even for a minor gap. A weak row records a finding with neither, which leaves no one to ask about it later. GOOD | Tabletop | 2026-03 | Activation and notification sequence | Internal audit | Notification to the regional scanning vendor took 40 minutes longer than planned | Ops Manager | 2026-04-15 | WEAK | Drill | sometime | everything | the team | some gaps | | (no observer independent of the team, no owner, no date) TRAP Letting an unmanned test, where someone else's staff run the recovery from backup tapes while your own team only watches, count as a sufficient test; or testing only the technical process and leaving out the people who would actually run it during a real disruption. -->
| Exercise Type | Date | What It Validates | Observer | Finding | Corrective Action Owner | Due Date ||---|---|---|---|---|---|---|| {{exercise_type}} | {{exercise_date}} | {{exercise_validates}} | {{exercise_observer}} | {{exercise_finding}} | {{corrective_action_owner}} | {{corrective_action_due_date}} |
## Plan Administration
<!-- WHAT Vendors and contractors this plan depends on, a dated revision history, and where a reachable copy of the plan is kept, including who holds the master copy. WHY FEMA names this cluster of sections directly: "Vendors & Contractors," "Revision History," and "Plan Distribution & Access," with its own distribution rule stated plainly: "A master copy of the document should be maintained by the business continuity team leader." The reason a plan needs a reachable, offline copy is sourced to the kind of event the plan exists to survive: a federal ransomware response guide asks that organizations "Ensure a hard copy of the plan and an offline version is available," and FEMA's continuity guidance asks organizations to "strongly consider multiple redundant media for storing their essential records." FINRA ties this section to an update trigger, not only a storage rule: "Each member must update its plan in the event of any material change to the member's operations, structure, business or location." Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Plan Administration). ASK Which vendors and contractors does this plan depend on, and what do they each actually provide? Is there a dated revision history? Where is a copy of this plan kept that does not depend on the systems the plan protects? Who holds the master copy? PRIORITY List vendors whose failure would stop a critical process, named in Business Impact Analysis Summary above, before ones that merely support it. ROW HINT A good vendor row names what the vendor actually provides and a contact method that still works if the vendor's own portal is down. A weak row is a name with no service and no backup contact method. A good revision row says what changed and which event caused it, such as an exercise finding, an activation, or a change in people or suppliers; a weak one says only "updated". GOOD | Regional scanning vendor | Scans and routes paper claims when the office is inaccessible | 24-hour support line, not only the vendor's ticketing portal | WEAK | Northline Scanning | (no service stated, and no contact method that works if their portal is down) TRAP Keeping the only copy of this plan on the same network or in the same facility the plan exists to recover. That is a stated precaution in the sources this bundle read, not a measured failure, but the logic holds either way: a plan reachable only through the systems it protects is unreachable exactly when it is needed. -->
| Vendor or Contractor | What They Provide | Contact Method ||---|---|---|| {{vendor_name}} | {{vendor_service}} | {{vendor_contact_method}} |
| Version | Date | Changed By | Summary of Change ||---|---|---|---|| {{revision_version}} | {{revision_date}} | {{revision_author}} | {{revision_summary}} |
**Master copy held by:** {{master_copy_holder}}**Where a reachable, offline copy of this plan is kept:** {{offline_copy_location}}
## Review Trigger
<!-- WHAT A named owner and the conditions that would make this plan itself wrong: a change in people, suppliers, systems, facilities or the business; a change to the Business Impact Analysis; an activation; or an exercise or test finding. Not a calendar date alone. WHY No source read for this bundle publishes a section with this name or job; it is required directly by this library's standing-standards family contract, and this bundle labels it as its own contribution rather than received practice. The conditions it names are drawn from sources that state a trigger without naming this section: NIST's narrative sets the general rule, "the plan should be reviewed for accuracy and completeness at an organization-defined frequency or whenever significant changes occur to any element of the plan," adding that "Certain elements, such as contact lists, will require more frequent reviews." The FFIEC ties a review to "any changes to key personnel, policies, procedures, facilities, equipment, outsourcing relationships, vendors, or other components that impact a critical business function." HMPPS ties a review to the same two events this section names, "annually or following an incident or exercise." Where a regulator or policy requires an annual exercise, it appears here as a floor under these conditions, not as the trigger itself. Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Review Trigger) and section 6, item 3. ASK Who is the single named owner of this section? What specific events would make you revise this plan, beyond a calendar date alone? PRIORITY List event-based conditions before any calendar floor, so the floor reads as a minimum under them rather than as the headline rule. ROW HINT A good row names a specific event, a named owner, and what they do about it. A weak row names only a calendar date. GOOD | A failed finding from a tabletop or drill | Business Continuity Team Lead | Investigate which existing procedure should have caught it before writing a new one | WEAK | Review this plan annually | The team | Update if needed | TRAP Treating an annual exercise that a regulator or internal policy requires as the trigger itself, rather than as a floor layered under the condition-based trigger above it. -->
| Condition That Would Make This Plan Wrong | Owner Who Notices | What They Do About It ||---|---|---|| {{review_trigger_condition}} | {{review_trigger_owner}} | {{review_trigger_action}} |
**Annual exercise required by policy or regulation, if any (a floor under the conditions above, not asubstitute for them):** {{annual_exercise_floor_note}}business-continuity-plan_template-full.md · ~7,800 tokens
---title: "{{plan_title}}"business_unit: "{{business_unit}}"owner: "{{owner}}"status: "{{status}}"last_updated: "{{date}}"doc_type: business-continuity-plansize: fullsource_template: business-continuity-plansource_template_version: 0.1.0---
<!--FULL BUSINESS CONTINUITY PLAN. Everything the lean variant carries, plus a Disaster Recovery sectionimmediately before Review Trigger: the recovery strategy for each critical system, backup and offsitestorage, recovery procedures, recovery order, restore verification, and the recovery test. Use it whenthis plan must carry disaster recovery itself, because no separate disaster recovery plan documents it.To go back to the load-bearing core, see business-continuity-plan_template-lean.md.
THIS VARIANT IS A STRICT SUPERSET OF THE LEAN ONE. The eight lean sections appear here in the same order,with the same headings and placeholders; full only inserts Disaster Recovery before Review Trigger. If youare growing from lean, add the new section and do not reorder or rename anything you already filled in.
THIS PLAN KEEPS BUSINESS PROCESSES GOING; ITS DISASTER RECOVERY SECTION RECOVERS THE SYSTEMS THOSEPROCESSES USE. NIST defines the plan as "The documentation of a predetermined set of instructions orprocedures that describe how an organization's mission/business processes will be sustained during andafter a significant disruption," and FEMA's Ready Business program ships it as a fill-in template.Disaster recovery is narrower: NIST scopes it to "an alternate facility." No source read ships a fill-intemplate titled as a standalone disaster recovery plan, so the Disaster Recovery section below is thislibrary's own grouping of California's technology recovery structure. Seebusiness-continuity-plan_companion.md section 1 and section 6, items 1 and 2.
THIS PLAN IS A STANDING INSTRUMENT, NOT A RECORD OF ONE DISRUPTION. A team executes it under timepressure once a disruption is declared or a drill begins. Update this file when the plan itself needs tochange, not every time it is activated. See business-continuity-plan_companion.md section 6, item 3.
WHAT THIS PLAN IS NOTIt is NOT a runbook, which is one operational procedure for a known, narrower situation; this plan can callout to one. It is NOT a production readiness review, which checks a service's fitness for a routinehandoff. It is NOT an incident postmortem, written after one event has resolved. It is NOT a risk register,which tracks events before they happen. And a high-availability design is not disaster recovery by itself.See business-continuity-plan_companion.md section 8.
HOW TO FILL THIS IN1. Read the comment under each heading: WHAT it wants, WHY it matters (with a pointer into business-continuity-plan_companion.md), guiding questions to ASK, a GOOD and a WEAK example, and the TRAP to avoid. For tables, PRIORITY explains the ordering rule and ROW HINT says what a good row contains.2. Replace each {{placeholder}} with your content. Fill Business Continuity Team and Roles and Business Impact Analysis Summary first; the rest, including Disaster Recovery, depends on knowing who runs this plan and which processes and systems it protects.3. If a section, or one row inside a table, does not apply, write "N/A" and the reason beside it rather than leaving it blank or deleting the section. An unexplained "N/A" and a skipped item look identical to a later reader.4. Before you ship it: self-grade against business-continuity-plan_guide.md, then DELETE every HTML comment. They are guidance, not content.-->
# {{plan_title}}
## Purpose and Scope
<!-- WHAT The business or unit this plan covers, what counts as a significant disruption in this plan's own scope, and where a separate disaster recovery plan lives if this variant's own Disaster Recovery section is not being used. WHY NIST defines the type this plan fills: "The documentation of a predetermined set of instructions or procedures that describe how an organization's mission/business processes will be sustained during and after a significant disruption." FEMA's Ready Business template lists "Information Technology Disaster Recovery Plan (if not included in the business continuity plan)" among the related plans a complete program keeps, which is the source for the pointer below: it stays available even in this variant, for an organization that maintains disaster recovery as its own separate document and does not need this plan's Disaster Recovery section at all. Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Purpose and Scope) and section 6, item 1. ASK What business or unit does this plan cover? What counts as a "significant disruption" here, in plain language? Will this plan's own Disaster Recovery section below carry the technology recovery detail, or does a separate disaster recovery plan already exist as its own document? If the latter, name it and say where it is kept. GOOD "This plan covers Meridian Freight's claims-processing division. A significant disruption is any event, a facility loss, an extended system outage, a regional disaster, that prevents claims processing for more than four hours. Technology recovery is carried in this plan's own Disaster Recovery section below, not in a separate document." WEAK "This is our business continuity plan." (no scope, no definition of a disruption, and no statement of whether disaster recovery lives in this document or elsewhere) TRAP Treating "disaster recovery plan" and "business continuity plan" as two names for one document. NIST and FEMA both treat disaster recovery as a narrower plan or section that may sit inside or beneath this one; neither source treats the two as equivalent. -->
{{continuity_scope_statement}}
**Separate disaster recovery plan, if one exists outside this document (otherwise, see the DisasterRecovery section below):** {{disaster_recovery_plan_pointer}}
## Business Continuity Team and Roles
<!-- WHAT The team roster by job title, the person with authority to activate this plan, a named successor, and any delegation that takes effect when normal channels of direction are disrupted. WHY FEMA's template carries its own "Business Continuity Organization" section naming the team. California's instructions ask that "job titles (rather than the names of individuals) should be used to assign responsibility as it lessens maintenance on procedures as personnel changes." NIST requires a single activation authority: "Only one individual should have this authority, and a successor should be clearly identified to assume that responsibility if necessary." FEMA's continuity guidance adds the delegation mechanism: "predetermined delegations of authority will take effect when normal channels of direction are disrupted and will terminate when these channels are reestablished," and asks that delegates also be named "by position title and not by name." Who actually writes, activates, and executes a plan like this is not settled across the sources this bundle read; name your own roles explicitly rather than assuming one pattern applies here. Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Business Continuity Team and Roles) and section 6, item 4. ASK What roles does this plan depend on, by job title? Who currently holds each one? Who has authority to activate this plan, and who is that person's named successor? Does a delegation clause need to state what happens if neither is reachable? PRIORITY List the activation authority and successor first, so a reader under time pressure finds the one answer they need before anything else on this page. ROW HINT A good row names a job title, not only a person, and states what that role actually does when this plan runs. A weak row names a person with no title and no stated backup. GOOD | Business Continuity Team Lead | Priya Shah, Director of Operations | Has authority to activate this plan; successor is the Deputy Director of Operations | WEAK | Team Lead | Priya | (no title, no successor, nothing to fall back on if Priya is unreachable) TRAP Naming individuals instead of job titles. California's own instructions give the reason directly: it "lessens maintenance on procedures as personnel changes." A plan that names only people goes stale the day one of them leaves. -->
| Role | Job Title and Current Holder | Responsibility ||---|---|---|| {{team_role}} | {{team_role_holder}} | {{team_role_responsibility}} |
**Authority to activate this plan:** {{activation_authority_title_and_holder}}**Named successor if the activation authority is unreachable:** {{activation_successor_title_and_holder}}**Delegation of authority, if normal channels of direction are disrupted:** {{delegation_of_authority_statement}}
## Business Impact Analysis Summary
<!-- WHAT Each critical business process, its Recovery Time Objective and Recovery Point Objective, and the systems and suppliers it depends on, carried over from a separate Business Impact Analysis. WHY NIST frames the analysis as upstream and prior: "The BIA is the first source for determining resiliency and contingency planning strategies," and its results "should be appropriately incorporated into" the plans that follow. FEMA's own template field reads "Insert results of Business Impact Analysis," and California requires that the recovery plan "must be derived from" the BIA. The two objectives are defined consistently across the sources this bundle read: a Recovery Time Objective is "the maximum acceptable delay between the interruption of service and restoration of service," and a Recovery Point Objective is "the maximum acceptable time after the last data recovery point." A supplier contract states the same obligation this table carries for suppliers: "Supplier shall ensure all Subcontractors have Business Continuity Programs and Disaster Recovery Plans sufficient to meet delivery of the Services during a Disruptive Event." Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Business Impact Analysis Summary). ASK For each critical process: what is its Recovery Time Objective and Recovery Point Objective, in plain units of time? What systems does it depend on? What suppliers does it depend on? PRIORITY List the process with the shortest Recovery Time Objective first, so the most time-critical process is visible without reading the whole table. ROW HINT A good row states Recovery Time Objective and Recovery Point Objective in concrete time units and names specific systems and suppliers, not "various." A weak row writes "TBD" or leaves the dependency columns blank. GOOD | Claims intake | 4 hours | 1 hour | Claims database; the regional scanning vendor | WEAK | Claims intake | Fast | N/A | (no units, and the dependency columns are not actually filled in) TRAP Deriving Recovery Time Objectives and Recovery Point Objectives from scratch in this table. This table carries a separate Business Impact Analysis's results; it is not the analysis itself. If you are deriving these figures for the first time, that work happens in a Business Impact Analysis, not here, and the per-system figures in this variant's Disaster Recovery section below must stay consistent with what this table already states. -->
| Critical Process | Recovery Time Objective | Recovery Point Objective | Systems and Suppliers Depended On ||---|---|---|---|| {{critical_process}} | {{process_rto}} | {{process_rpo}} | {{process_dependencies}} |
## Continuity Strategies and Manual Workarounds
<!-- WHAT How each critical process keeps running without the systems it normally depends on, and at what recovery level, chosen against cost. WHY FEMA's template names this as its own section, "Manual Workarounds." The cost tradeoff that governs how far to invest in a workaround is stated most explicitly by AWS's own recovery strategies, ordered "in increasing order of cost and complexity, and decreasing order of RTO and RPO," with an explicit warning: "You should avoid implementing a strategy that is more stringent than it needs to be, as this incurs unnecessary costs." Google states the same tradeoff for the process as a whole: "Typically, the smaller your RTO and RPO values are... the more your application will cost to run." A high-availability technical design, by itself, is not a continuity strategy for the business process it supports: "HA doesn't entirely overlap with DR, but it's often necessary to take HA into account when you're thinking about RTO and RPO values." Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Continuity Strategies and Manual Workarounds). ASK For each critical process, what do staff actually do by hand when the normal system is unavailable? What recovery level was chosen, and why, given what it costs against how fast the process actually needs to recover? PRIORITY List the process with the shortest Recovery Time Objective, from Business Impact Analysis Summary above, first, since it is first in line for investment. ROW HINT A good row states what staff actually do, step by step, not that a workaround "exists." A weak row asserts a workaround with no description of the steps. GOOD | Claims intake | Staff log claims to a shared paper form and a backup laptop with offline storage, then enter them into the database once it is back | Manual, low cost, matches the process's own four-hour Recovery Time Objective | WEAK | Claims intake | Manual workaround exists | (no description of what staff actually do, and no stated reason for the recovery level chosen) TRAP Choosing a workaround faster than the process actually needs. AWS's own warning applies here even though it was written about technology recovery: a workaround that beats its process's own Recovery Time Objective is also one that costs more than it needs to. -->
| Critical Process | Manual Workaround | Recovery Level Chosen, and Why ||---|---|---|| {{workaround_process}} | {{workaround_description}} | {{workaround_recovery_level_rationale}} |
## Activation and Incident Management
<!-- WHAT Stated activation criteria, who declares an activation, who is told and in what order, and the condition for standing the plan down. WHY NIST requires that activation criteria be stated in advance and tied to a designated authority: "The designated authority... has the authority to activate the contingency plan," triggered when "one or more of the activation criteria for that system are met." California asks the plan to identify "individual(s) responsible for making the TRP activation decision." HMPPS separates who activates from who writes the plan: "A member of the nominated Business Continuity team will normally activate and stand down the plans." Where regulation requires it, notice extends past an organization's own staff: FINRA binds its member firms to "disclose to its customers how its business continuity plan addresses the possibility of a future significant business disruption." Deactivation is its own named step, not an afterthought: NIST's reconstitution phase "consists of two major activities: validating successful reconstitution and deactivation of the plan." Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Activation and Incident Management). ASK What specific conditions trigger an activation? Who declares one? Who is notified, in what order, and by what method, including anyone outside the organization you owe notice to? What condition stands the plan down? PRIORITY List notifications in the actual order they go out, so the first call someone makes during a real disruption is the first row on this page. ROW HINT A good row names a specific role or entity and a method that still works if the usual channel is down. A weak row says "notify everyone" with no order and no method. GOOD | 1 | Business Continuity Team Lead | Paged directly; confirms the activation decision | WEAK | | Everyone | (no order, no method, and no way to tell who actually gets called first) TRAP Treating a regulatory or customer notification duty as universal. FINRA's duty binds its own member firms specifically; fold a notification line like it in only where your own organization actually owes that notice. -->
**Activation criteria:** {{activation_criteria}}**Who declares an activation:** {{activation_decision_maker}}
| Order | Who Is Notified | Method ||---|---|---|| {{notification_order}} | {{notification_recipient}} | {{notification_method}} |
**Condition for standing this plan down:** {{stand_down_condition}}
## Training, Testing and Exercising
<!-- WHAT The exercise schedule and type, what each one validates, who observes it, and each finding as an owned, dated corrective action. WHY The FFIEC names four test types, from lightest to heaviest: "Orientation/Walk-through," "Tabletop/Mini-drill," "Functional Testing," and "Full-scale Testing," and names a specific insufficient test: "Unmanned recovery testing, where back-up tapes are sent to the recovery site to be run by service provider employees, is not a sufficient test." The same source asks that "the audit department or other independent party should directly observe tests." A test is judged against named criteria, not a bare pass or fail: Google names three, "data integrity, recovery time objective (RTO), and recovery point objective (RPO)." Microsoft adds that a test must exercise the people, not only the systems: "Include all of the human processes in your drills, and don't just focus on the technical processes." Findings become tracked, owned work: FEMA's continuity guidance names a standing "Corrective Action Program (CAP)," where "Deficiencies, actions to correct them, and a timeline for remedy are documented," and HMPPS requires that "Debrief sessions must be held following an exercise and lessons learnt log completed." A practitioner argues directly that these activities carry no pass or fail judgement at all, "these activities are an exercise, not a test"; this bundle keeps both words in this section's own title rather than choosing a side. A failed restore test from this variant's own Disaster Recovery section below is the same kind of finding as any other exercise finding here. Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Training, Testing and Exercising) and section 6, item 6. ASK What type of exercise is scheduled? What is it meant to validate? Who observes it, independently of the team being exercised? For every finding, who owns the fix, and by what date? PRIORITY List the most recent exercise first, so a reader can see immediately whether this plan has actually been tested recently. ROW HINT A good row's finding carries a named owner and a due date, even for a minor gap. A weak row records a finding with neither, which leaves no one to ask about it later. GOOD | Tabletop | 2026-03 | Activation and notification sequence | Internal audit | Notification to the regional scanning vendor took 40 minutes longer than planned | Ops Manager | 2026-04-15 | WEAK | Drill | sometime | everything | the team | some gaps | | (no observer independent of the team, no owner, no date) TRAP Letting an unmanned test, where someone else's staff run the recovery from backup tapes while your own team only watches, count as a sufficient test; or testing only the technical process and leaving out the people who would actually run it during a real disruption. -->
| Exercise Type | Date | What It Validates | Observer | Finding | Corrective Action Owner | Due Date ||---|---|---|---|---|---|---|| {{exercise_type}} | {{exercise_date}} | {{exercise_validates}} | {{exercise_observer}} | {{exercise_finding}} | {{corrective_action_owner}} | {{corrective_action_due_date}} |
## Plan Administration
<!-- WHAT Vendors and contractors this plan depends on, a dated revision history, and where a reachable copy of the plan is kept, including who holds the master copy. WHY FEMA names this cluster of sections directly: "Vendors & Contractors," "Revision History," and "Plan Distribution & Access," with its own distribution rule stated plainly: "A master copy of the document should be maintained by the business continuity team leader." The reason a plan needs a reachable, offline copy is sourced to the kind of event the plan exists to survive: a federal ransomware response guide asks that organizations "Ensure a hard copy of the plan and an offline version is available," and FEMA's continuity guidance asks organizations to "strongly consider multiple redundant media for storing their essential records." FINRA ties this section to an update trigger, not only a storage rule: "Each member must update its plan in the event of any material change to the member's operations, structure, business or location." Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Plan Administration). ASK Which vendors and contractors does this plan depend on, and what do they each actually provide? Is there a dated revision history? Where is a copy of this plan kept that does not depend on the systems the plan protects? Who holds the master copy? PRIORITY List vendors whose failure would stop a critical process, named in Business Impact Analysis Summary above, before ones that merely support it. ROW HINT A good vendor row names what the vendor actually provides and a contact method that still works if the vendor's own portal is down. A weak row is a name with no service and no backup contact method. A good revision row says what changed and which event caused it, such as an exercise finding, an activation, or a change in people or suppliers; a weak one says only "updated". GOOD | Regional scanning vendor | Scans and routes paper claims when the office is inaccessible | 24-hour support line, not only the vendor's ticketing portal | WEAK | Northline Scanning | (no service stated, and no contact method that works if their portal is down) TRAP Keeping the only copy of this plan on the same network or in the same facility the plan exists to recover. That is a stated precaution in the sources this bundle read, not a measured failure, but the logic holds either way: a plan reachable only through the systems it protects is unreachable exactly when it is needed. -->
| Vendor or Contractor | What They Provide | Contact Method ||---|---|---|| {{vendor_name}} | {{vendor_service}} | {{vendor_contact_method}} |
| Version | Date | Changed By | Summary of Change ||---|---|---|---|| {{revision_version}} | {{revision_date}} | {{revision_author}} | {{revision_summary}} |
**Master copy held by:** {{master_copy_holder}}**Where a reachable, offline copy of this plan is kept:** {{offline_copy_location}}
## Disaster Recovery
<!-- WHAT The recovery strategy for each critical system; backup and offsite storage, including at least one offline copy; recovery procedures written so another trained person could follow them; the order of recovery across dependent systems; how restored data is verified before service resumes; and the recovery test. WHY AWS orders recovery strategies "in increasing order of cost and complexity, and decreasing order of RTO and RPO," and warns against "a strategy that is more stringent than it needs to be." A joint government ransomware guide requires offline backups, because "many ransomware variants attempt to find and subsequently delete or encrypt accessible backups," and asks that immutable storage be used "with caution." Google's planning guide notes that high availability "doesn't entirely overlap with DR." NIST and California both ask for procedures that someone other than their author could follow. Recovery order matters because "After an outage, you can end up with a deadlock: two systems that each can't start until the other is running and providing some functionality." Google judges a restore test on "data integrity, recovery time objective (RTO), and recovery point objective (RPO)." Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Disaster Recovery). ASK For each critical system: what is its recovery strategy? Where do its backups live, including at least one offline copy? What order must it recover in, relative to the systems it depends on? How will you confirm the restored data is actually usable before declaring the system back in service? PRIORITY List systems in the actual order they must recover, upstream dependencies first, not alphabetically, so the table itself shows the sequence a real recovery follows. ROW HINT A good row names a specific, checkable recovery strategy, not "we will restore from backup" with no further detail, and states its dependency order explicitly. A weak row has backups with no stated test, no verification criteria, and no recovery order. GOOD | Claims database | Warm standby in a second region; restore from the last automated snapshot if standby fails | Recovers after the authentication service it depends on | WEAK | Claims database | We have backups | (no recovery order, no verification criteria, and no record of when it was last tested) TRAP Treating backups as a recovery. Having backups is not the same claim as having one, and a practitioner states the gap directly: "If you have backups, but you haven't tested restoring them, you don't really have backups." -->
| Critical System | Recovery Strategy | Recovery Order (What It Depends On) ||---|---|---|| {{dr_system_name}} | {{dr_recovery_strategy}} | {{dr_recovery_order}} |
**Backup and offsite storage, including at least one offline copy:** {{backup_and_offsite_storage}}**Recovery procedures (written so another trained person could follow them):** {{recovery_procedures_reference}}**Restore verification criteria (data integrity, Recovery Time Objective, Recovery Point Objective):** {{restore_verification_criteria}}**Recovery test schedule and type, and when it was last run:** {{recovery_test_schedule}}
## Review Trigger
<!-- WHAT A named owner and the conditions that would make this plan itself wrong: a change in people, suppliers, systems, facilities or the business; a change to the Business Impact Analysis; an activation; or an exercise or test finding, including a failed restore test from the Disaster Recovery section above. Not a calendar date alone. WHY No source read for this bundle publishes a section with this name or job; it is required directly by this library's standing-standards family contract, and this bundle labels it as its own contribution rather than received practice. The conditions it names are drawn from sources that state a trigger without naming this section: NIST's narrative sets the general rule, "the plan should be reviewed for accuracy and completeness at an organization-defined frequency or whenever significant changes occur to any element of the plan," adding that "Certain elements, such as contact lists, will require more frequent reviews." The FFIEC ties a review to "any changes to key personnel, policies, procedures, facilities, equipment, outsourcing relationships, vendors, or other components that impact a critical business function." HMPPS ties a review to the same two events this section names, "annually or following an incident or exercise." Where a regulator or policy requires an annual exercise, it appears here as a floor under these conditions, not as the trigger itself. Deep dive: business-continuity-plan_companion.md section 3 (Anatomy > Review Trigger) and section 6, item 3. ASK Who is the single named owner of this section? What specific events would make you revise this plan, beyond a calendar date alone, including a failed finding from the Disaster Recovery section's own recovery test above? PRIORITY List event-based conditions before any calendar floor, so the floor reads as a minimum under them rather than as the headline rule. ROW HINT A good row names a specific event, a named owner, and what they do about it. A weak row names only a calendar date. GOOD | A failed recovery test from the Disaster Recovery section | Business Continuity Team Lead | Investigate which existing procedure should have caught it before writing a new one | WEAK | Review this plan annually | The team | Update if needed | TRAP Treating an annual exercise that a regulator or internal policy requires as the trigger itself, rather than as a floor layered under the condition-based trigger above it. -->
| Condition That Would Make This Plan Wrong | Owner Who Notices | What They Do About It ||---|---|---|| {{review_trigger_condition}} | {{review_trigger_owner}} | {{review_trigger_action}} |
**Annual exercise required by policy or regulation, if any (a floor under the conditions above, not asubstitute for them):** {{annual_exercise_floor_note}}business-continuity-plan_example.md
---title: "Platform Team Business Continuity Plan"business_unit: "Platform, Acme Analytics"owner: "Dana Osei (Staff Engineer, Platform)"status: "active"last_updated: "2026-09-15"doc_type: business-continuity-plansize: fullsource_template: business-continuity-plansource_template_version: 0.1.0---
> **Worked example.** A filled `business-continuity-plan`, full variant, for Acme Analytics' Platform> team: the standing plan that keeps the shared main Postgres database available through a significant> disruption, the instrument the> [`production-readiness-review`](../production-readiness-review/production-readiness-review_example.md)> example names without showing, when it marks its own "Data and backup, disaster recovery" criterion Not> Applicable on the ground that this plan already covers it. Per the `standing-standards` family> contract, the chaining stays loose by design: this is the standing instrument a team agrees once and> consults repeatedly, not a record written for one disruption. It is dated **2026-09-15**, after every> document it cites: the [`sdd`](../sdd/sdd_example.md) and> [`definition-of-done`](../definition-of-done/definition-of-done_example.md) examples' Saved Views work,> the [`definition-of-ready`](../definition-of-ready/definition-of-ready_example.md) example, the> [`runbook`](../runbook/runbook_example.md) example's entitlement-audit incident, and the> [`production-readiness-review`](../production-readiness-review/production-readiness-review_example.md)> example's decision, on 2026-08-22, that Acme Analytics' new SRE function takes standing on-call> ownership of dashboard-service. Everything cited below had already happened. This plan's own first> version predates all of it: its revision history carries an entry from 2026-03-02, months before the> review that first names it. Read it alongside> [`business-continuity-plan_guide.md`](business-continuity-plan_guide.md), the rubric it was graded> against. All names not already established in the library's Acme Analytics thread, and every figure,> date, and status not otherwise cited, are illustrative. That includes billing-service and> notifications-service, the two further dependents of the shared database this plan names: no earlier> example enumerates the database's dependents beyond `saved_view`, so this one introduces them.
# Platform Team Business Continuity Plan
## Purpose and Scope
This plan covers Acme Analytics' Platform team and the one piece of shared infrastructure every otherteam's service ultimately sits on top of: the shared main Postgres database that dashboard-service's`saved_view` table, among other teams' tables, lives in. A significant disruption, for this plan, is anyevent that leaves that database unreachable, or returning errors, for any dependent service for more thanten minutes, or that makes the cloud region it runs in unavailable. A disruption to one dependent team'sown application code, with the database itself healthy, is that team's own incident and does not activatethis plan.
**Separate disaster recovery plan, if one exists outside this document (otherwise, see the DisasterRecovery section below):** None. No separate disaster recovery plan exists for the shared database; thisvariant's own Disaster Recovery section below carries that detail, because Platform decided, when thisplan was first written, that the database's recovery strategy belongs next to the business impact itprotects rather than in a document of its own.
## Business Continuity Team and Roles
| Role | Job Title and Current Holder | Responsibility ||---|---|---|| Business Continuity Team Lead | Dana Osei, Staff Engineer, Platform | Owns this plan, has authority to activate it, and is the single person every dependent team escalates a shared-database disruption to || Database Reliability Lead | Rosa Chen, Senior Engineer, Platform | Executes the recovery procedures named in Disaster Recovery below, and owns the corrective actions this plan's own exercises produce || Dependent-Team Liaison | Whoever is currently on call for the affected dependent service (see Activation and Incident Management for who that includes today) | Confirms to the Business Continuity Team Lead which of their own team's critical processes are actually affected, so this roster does not have to name every dependent team's engineer in advance |
**Authority to activate this plan:** Dana Osei, Staff Engineer, Platform**Named successor if the activation authority is unreachable:** Rosa Chen, Senior Engineer, Platform**Delegation of authority, if normal channels of direction are disrupted:** If neither Dana Osei nor RosaChen is reachable within fifteen minutes (illustrative) of a declared disruption, the Platform engineercurrently on call for the shared database may activate this plan directly, and must notify Dana Osei andRosa Chen as soon as normal channels of direction are reestablished.
## Business Impact Analysis Summary
| Critical Process | Recovery Time Objective | Recovery Point Objective | Systems and Suppliers Depended On ||---|---|---|---|| Serving reads and writes to the shared primary Postgres database | 30 minutes (illustrative) | 5 minutes (illustrative) | dashboard-service (Reporting, see [`sdd_example.md`](../sdd/sdd_example.md)); billing-service (Finance Engineering, illustrative); notifications-service (Growth, illustrative) || Platform's internal admin console, used to triage and declare a disruption | 8 hours (illustrative) | Not applicable; the console holds no data of its own | The shared primary Postgres database above, for its own audit log only |
## Continuity Strategies and Manual Workarounds
| Critical Process | Manual Workaround | Recovery Level Chosen, and Why ||---|---|---|| Serving reads and writes to the shared primary Postgres database | Platform posts a status-page entry asking every dependent team to degrade to its own read-only or cached behavior while Platform executes the Disaster Recovery section's recovery strategy; Platform does not attempt to serve writes from anywhere other than the recovered database | Warm standby, not a faster multi-region active-active setup: the database's own Recovery Time Objective above is thirty minutes (illustrative), and a warm standby already meets it without the added cost and operational complexity an active-active setup would add || Platform's internal admin console | Platform engineers on the Database Reliability rotation operate directly against the database through command-line tooling, bypassing the console entirely | Manual: the console is a convenience for declaring and tracking a disruption, not part of serving any dependent team, so an eight-hour (illustrative) objective leaves room to work without it, and a second console would buy nothing the command line does not already give |
## Activation and Incident Management
**Activation criteria:** The shared primary Postgres database is unreachable, or returning errors, forany dependent service for more than ten minutes, or the cloud region it runs in is declared unavailable.**Who declares an activation:** Dana Osei, Staff Engineer, Platform, or Rosa Chen as named successor.
| Order | Who Is Notified | Method ||---|---|---|| 1 | The Platform engineer currently on call for the shared database | Paged directly through Platform's own on-call tool || 2 | Dependent-team liaisons for dashboard-service, billing-service and notifications-service | Paged through each team's own on-call tool, plus the shared status page, so a dependent team still learns of the disruption if its own paging channel is itself the thing that is down || 3 | Acme Analytics' SRE on-call rotation (added in version 1.2.0 on 2026-08-25, after the 2026-08-22 production readiness review decided SRE would carry dashboard-service's own pager; dashboard-service depends on the database this plan protects, so SRE now needs the same notice a dependent team gets) | Paged through the same on-call tool dashboard-service's own on-call already uses || 4 | Acme Analytics' internal status page, for any team not named above that also reads from the shared database | A status-page entry, updated at each stage of recovery |
**Condition for standing this plan down:** The Database Reliability Lead confirms the database is servingreads and writes within the Recovery Time and Recovery Point Objectives stated in Business ImpactAnalysis Summary above, and thirty minutes (illustrative) pass with no recurrence.
## Training, Testing and Exercising
| Exercise Type | Date | What It Validates | Observer | Finding | Corrective Action Owner | Due Date ||---|---|---|---|---|---|---|| Tabletop | 2026-09-05 (illustrative) | The delegation of authority in Business Continuity Team and Roles above: neither Dana Osei nor Rosa Chen can be reached, so the Platform engineer on call for the shared database must activate this plan and begin recovery alone | Ines Halvorsen (Engineer, SRE), observing independently of Platform, the team being exercised: dashboard-service, which SRE now carries the pager for, depends on the database this plan protects, which is why SRE asked to observe this exercise rather than only receive its finding afterward | The on-call engineer could activate the plan but lacked the hosting provider permission needed to promote the standby replica, so recovery would have waited until one of the two leads was reached | Rosa Chen | 2026-10-01 (illustrative) || Full-scale failover test | 2026-06-10 (illustrative) | An actual failover of the shared primary Postgres database to its standby region, judged on data integrity, Recovery Time Objective and Recovery Point Objective | Acme Analytics' internal audit function | Failover itself completed inside the stated Recovery Time Objective, but verifying the restored data's integrity on the standby copy took long enough to come close to breaching the Recovery Point Objective | Rosa Chen | 2026-06-24 (illustrative; closed in version 1.1.0 by writing the restore verification criteria now stated in Disaster Recovery below) |
## Plan Administration
| Vendor or Contractor | What They Provide | Contact Method ||---|---|---|| Platform's managed Postgres hosting provider | Hosts the primary database and its standby replica across two regions | The enterprise support phone line printed in the on-call binder. The provider's web console is assumed unreachable during a regional outage, which is exactly when this line is needed || The offsite backup storage vendor | Stores the nightly encrypted snapshot named in Disaster Recovery below, in a third region separate from both the primary and standby database regions | A named account representative's direct line, confirmed quarterly rather than assumed current |
| Version | Date | Changed By | Summary of Change ||---|---|---|---|| 1.0.0 | 2026-03-02 (illustrative) | Dana Osei | Initial plan: the shared primary Postgres database's business impact analysis, activation criteria and recovery strategy || 1.1.0 | 2026-06-24 (illustrative) | Rosa Chen | Added the restore verification criteria to Disaster Recovery, closing the 2026-06-10 failover test's finding || 1.2.0 | 2026-08-25 | Dana Osei | Added SRE as a notified stakeholder in Activation and Incident Management, following the 2026-08-22 production readiness review's decision that SRE takes standing on-call ownership of dashboard-service || 1.3.0 | 2026-09-15 | Dana Osei | Recorded the 2026-09-05 tabletop exercise and its finding in Training, Testing and Exercising, and the trigger that watches for its recurrence in Review Trigger |
**Master copy held by:** Dana Osei, Business Continuity Team Lead.**Where a reachable, offline copy of this plan is kept:** A printed copy in Platform's on-call binder,plus an offline PDF on the laptop issued to whoever currently holds the Platform on-call pager, neither ofwhich depends on the database this plan exists to recover.
## Disaster Recovery
| Critical System | Recovery Strategy | Recovery Order (What It Depends On) ||---|---|---|| Shared primary Postgres database | An existing read replica in the standby region is promoted to primary; if promotion itself fails, a fresh instance is restored from the most recent automated snapshot named in the backup statement below | Recovers first. Every dependent system named in Business Impact Analysis Summary above, dashboard-service, billing-service and notifications-service, waits on this database before attempting its own recovery |
**Backup and offsite storage, including at least one offline copy:** Continuous write-ahead log shippingkeeps the standby region's replica current to within the Recovery Point Objective stated above; a nightlyencrypted snapshot is additionally copied to offline storage in a third region, the one copy the hostingvendor's own production credentials cannot reach.**Recovery procedures (written so another trained person could follow them):** The Database ReliabilityLead's own step-by-step failover procedure, kept with the vendor contact information named in PlanAdministration above, written so a trained database engineer from outside Platform could carry it out.**Restore verification criteria (data integrity, Recovery Time Objective, Recovery Point Objective):** Apromoted or restored database is not declared back in service until a Database Reliability engineerconfirms three things: the row counts on the three largest tables, including dashboard-service's own`saved_view`, match the last known-good count within an explainable delta; write latency returns to itsnormal range; and any data actually lost stays inside the five-minute Recovery Point Objective statedabove. This criterion exists because the 2026-06-10 exercise above found the failover itself fast but itsrestore verification slow enough to risk breaching that objective.**Recovery test schedule and type, and when it was last run:** A full-scale failover test at least once ayear, last run 2026-06-10 (see Training, Testing and Exercising above), with a tabletop exercise inbetween.
## Review Trigger
| Condition That Would Make This Plan Wrong | Owner Who Notices | What They Do About It ||---|---|---|| A service begins depending on the shared primary Postgres database, or an existing dependent stops using it | Dana Osei | Update Business Impact Analysis Summary and Continuity Strategies and Manual Workarounds with the added or removed dependency before the next scheduled exercise, not only at the next annual review || The 2026-09-05 tabletop's finding above recurs: the next activation or exercise still finds the on-call engineer unable to promote the standby without one of the two leads | Dana Osei | Treat it as the corrective action from that finding not actually having worked, and escalate directly to the Database Reliability Lead, rather than opening it again as a new finding || The hosting provider changes the replication, snapshot or region arrangement the Disaster Recovery section relies on | Rosa Chen | Re-run the failover test against the changed arrangement before the old procedure is trusted again, and update the backup and offsite storage statement to match what is actually running |
**Annual exercise required by policy or regulation, if any (a floor under the conditions above, not asubstitute for them):** Platform runs the full-scale failover test described in Disaster Recovery atleast once a year as this plan's own floor under the conditions above, not as a substitute for them. The2026-09-05 tabletop is additional exercise, not the annual test itself.Provenance
Section titled “Provenance”The reasoning, the history and every source, in the repository:
- Companion - the long-form argument: why these sections, where the sources disagree, and what the bundle refuses to claim
- History - what changed in this bundle, and when
- Research log - every source consulted, with what each one actually supports
- Catalog metadata - the machine-readable record this page is generated from
Catalog record: 9 sections across 1 format(s).