Your shared documents disappear before a delivery deadline.
Who decides which work continues? Where does the team record new instructions? Who can tell a customer that the deadline has changed?
A business continuity tabletop exercise puts those decisions on the table before an outage puts them on the clock.
For a small service team, start with one unavailable system and one business outcome. Talk through the response. Record the missing evidence. Give each unresolved decision an owner.
The exercise below is a proposed discussion format. Its company, documents, deadlines, and scenario updates are fictional. Nobody disables a service, restores a file, or sends a message.
Define what the exercise can establish
NIST SP 800-34 Revision 1 describes a tabletop as a facilitated discussion of roles and responses. It separates that discussion from functional exercises and technical testing. The guide addresses federal information systems. This small-team worksheet adapts its exercise concepts; it creates no mandatory business control.
Use three separate evidence labels:
| Activity | Evidence it can produce | Question still open |
|---|---|---|
| Tabletop discussion | Decisions, assumptions, missing authority, proposed actions | Can people perform the actions? |
| Technical recovery test | Observed results for a defined system and test scope | Can the business resume its required work? |
| Business workflow validation | Observed completion of selected work after recovery | What remains untested? |
A participant saying they can access an offline copy remains a statement. The observer records it as unverified until a separate check supplies evidence.
Keep that distinction visible throughout the session. Agreement around the table cannot establish that a backup is readable or a permission works.
Choose one outcome and bound the room
Use an existing runbook and dependency map as inputs. This exercise rehearses decisions around them. It does not rebuild the inventory.
For the fictional team, the outcome is a reviewed delivery package ready for an afternoon customer meeting. The shared-document service becomes unavailable during preparation.
The scenario leaves email and the approved internal discussion channel available. It says nothing about the outage cause. Participants must avoid inventing a cause or a recovery estimate.
Write a short exercise card before starting:
- Outcome under discussion: one delivery package.
- System assumed unavailable: the shared-document workspace.
- Available inputs: a fictional task list, runbook extract, and document version notes.
- Decisions in scope: priorities, authority, communication, and temporary working arrangements.
- Actions outside scope: production changes, real messages, account access, and recovery execution.
Set the session length to fit the team. A proposed first run uses ten minutes for setup, thirty for scenario discussion, and twenty for review. These are agenda choices, not recovery targets.
Assign roles before presenting the outage
The facilitator presents facts, controls the fictional timeline, and asks questions. They separate scenario information from participant assumptions.
The decision owner chooses business priorities and records provisional decisions. The role needs the authority to discuss delayed work and escalation. It does not grant new operational permissions.
The observer records decisions, evidence references, uncertainty, and unanswered questions. They capture what the team says without turning a confident answer into a verified fact.
Participants represent the work: delivery, administration, technical support, and customer communication where relevant. A small team can combine roles. Keep observation separate from decision ownership when staffing permits.
AWS guidance on resilience game days includes business stakeholders, advance notice, and post-exercise review. Its operational game days have broader execution scope. This tabletop borrows the participation and follow-through ideas without introducing faults.
Before continuing, state the stop conditions. Pause for a real incident, accidental disclosure, an attempted production change, or confusion between exercise messages and real instructions.
If someone needs a live check to answer a question, record a follow-up. Do not turn the meeting into an improvised production test.
Establish the fictional starting position
Read this setup aloud:
Exercise only. It is 09:00 at a fictional service company. A delivery package needs review before a 15:00 customer meeting. Shared documents stop loading. The cause and recovery time are unknown. Email and the internal discussion channel remain available. No actual system is affected.
Give the team a fictional work queue with three items: the delivery package, an internal planning note, and a routine archive task.
Ask which item receives attention first and why. Require an explanation tied to a business commitment or dependency. Seniority and whoever speaks first are poor substitutes for a stated priority.
Then ask where the decision record lives during the assumed outage. If the proposed location depends on the unavailable service, flag the contradiction.
This is the opening condition. The facilitator now introduces exactly three scenario updates.
Update one: the available copy may be stale
At fictional time 09:20, a participant reports having yesterday's downloaded package. Another participant remembers an important revision this morning. Neither can establish the current approved version.
Ask the team:
- Which preparation tasks can continue without changing the package?
- What evidence would establish the approved revision?
- Who can authorize use of an older version?
- What work must pause while that decision remains unresolved?
- Where will temporary notes record their source and uncertainty?
The observer records access uncertainty separately from version uncertainty. Having a copy answers neither whether it is current nor whether its use is permitted.
A useful provisional decision might be to prepare a list of questions from the older copy while holding all customer delivery. Label that as the fictional team's choice. It is not a universal response to document loss.
If participants disagree, preserve both options and name the missing decision authority. The disagreement is an exercise finding, not a reason to smooth the notes.
Update two: the approver is unavailable
At fictional time 10:15, the usual delivery approver cannot be reached within the scenario. A customer asks whether the afternoon meeting still stands. No service restoration estimate is available.
Ask who has delegated authority and where that delegation is recorded. If the answer is unknown, keep it unknown. Familiarity with the customer does not establish approval rights.
Draft the content of a possible update inside the exercise notes. Identify the audience, sender role, facts known, next update point, and approval requirement. Do not enter recipients or send it.
For example, the proposed message can acknowledge an access problem affecting preparation and promise a further status update at a chosen time. It should avoid an unsupported restoration estimate.
Now challenge the communication decision. Does it disclose only what the audience needs? Does it imply completed review? Who owns the next update if the first sender becomes unavailable?
The asynchronous communication guide provides wider context for written team decisions. This exercise narrows that work to an outage with uncertain facts.
Update three: access returns with unresolved edits
At fictional time 12:30, documents load again. Two people have separate temporary notes. One has marked the delivery package ready in their own copy. The shared workspace does not show that approval.
The temptation is to declare recovery because the screen works. Keep the discussion on the business outcome.
Ask which version becomes authoritative, who compares the notes, and who confirms review. Identify whether any temporary work creates duplicate tasks or conflicting instructions.
Then ask what must be true before the team resumes normal handling. Possible conditions include reconciled revisions, verified access for the assigned reviewer, and a recorded delivery decision.
Those conditions remain proposed until a separate check observes them. The tabletop ends with a decision path, not a claim that the package has been recovered.
Record which routine work can resume immediately and which work stays held. Restoration of one service does not settle every deferred business decision.
Keep the decision record concrete
Use one row per decision. Avoid paragraphs that blend a fact, assumption, and instruction into the same note.
| Record field | What belongs in it |
|---|---|
| Scenario time | Fictional time and update number |
| Business priority | Work protected or deferred, with the reason |
| Evidence | Scenario card, runbook section, or unverified participant statement |
| Decision authority | Named role and delegation reference, or unknown |
| Communication | Audience, draft purpose, approver, next update point |
| Manual alternative | Proposed method, permitted inputs, limits, reconciliation need |
| Open question | Missing access, unclear ownership, or unresolved dependency |
Keep actual credentials, personal information, and customer files out of the worksheet. A role and document reference are enough for this fictional rehearsal.
If the team proposes copying material into another tool, ask who permits that destination. Record the data boundary before discussing convenience.
An operations automation audit can identify workflows for later inspection. Here, capture only the dependencies that block a scenario decision.
Turn the review into an action register
Give every participant a chance to name one assumption the exercise exposed. Compare those observations with the observer's notes before assigning fixes.
NIST's guide includes after-action reporting and plan updates. Use that principle to produce a short record with evidence and closure conditions.
Copy this blank worksheet for each finding:
| Field | Entry |
|---|---|
| Finding identifier | |
| Scenario update and decision affected | |
| Observed gap | |
| Supporting note or document reference | |
| What remains unverified | |
| Assigned owner | |
| Due date | |
| Proposed corrective action | |
| Evidence required for closure | |
| Retest condition and scope | |
| Reviewer and final disposition |
For the fictional stale-copy problem, the action might clarify where approved revisions are identified. Closure would require the revised instruction and a permitted check by another reader.
Updating a runbook is one result. Demonstrating that the replacement instruction works is another. Keep separate evidence for each.
Use the SOP documentation guide when converting an accepted decision into maintained instructions. Avoid adding every speculative option to the operating procedure.
Schedule the next question
Choose the next exercise from an unresolved finding. An absent approver may justify another discussion. Uncertain file recovery requires a separately authorized restore test.
Decide when the team revisits this scenario. A changed approval structure, replaced document service, or revised recovery process can provide a trigger. Choose the cadence according to the business.
Retain the exercise card, decision record, and action register together. Date them and identify which runbook versions participants used.
The useful output is specific: the team knows which decisions remain unsupported and who owns the next check.
Close the gaps. Rehearse again.
Sources
- NIST SP 800-34 Revision 1, updated publication: exercise types, recovery planning, and after-action follow-through.
- AWS Well-Architected: conduct game days regularly: stakeholder participation and review of exercise findings.