Risk Assessment

Safe Trial Risk Assessment & Rollback Plan

An interactive risk-assessment worksheet and rollback-planning form for low-risk pilots and trials. Capture trial scope, impacted systems and data, potential harms, detection signals and thresholds, mitigation controls, explicit rollback triggers and step-by-step rollback actions, consent and communications, decision criteria, and approvals. Includes help text and short examples to guide thresholds and monitoring.

Interactive Tool

Safe Trial Risk Assessment & Rollback Plan

Use this guided worksheet to document a low-risk pilot or trial so your team can learn quickly while minimizing harm, legal exposure, and reputational risk. Fill each section with realistic, testable detection signals and explicit rollback triggers. The form saves your responses so you can review, update, and export them later.

Quick guidance: Be concrete. Specify systems, data, thresholds, owners, notification lists and exact rollback steps. Where you see an example in HelpText, adapt it to your context—do not copy verbatim without checking local regulations or contracts.

A short descriptive name for this pilot (team-friendly).
Person or role accountable for the trial (who can call rollback).
YYYY-MM-DD (approximate if unknown).
Describe functionality, user segments, geography, duration, limits and any excluded items. Example: 'A/B test of new checkout flow for 5% of US web traffic for 2 weeks.'
List systems, integrations, data types (e.g., PII, PHI, payment data), third parties and subsystems that could be affected.
Check all that apply.
Describe possible customer, business, compliance, safety, or reputational harms. Be concrete (e.g., 'incorrect billing for 0.5% of transactions' or 'login failures preventing 2% of active sessions').
A quick qualitative assessment to prioritize monitoring and controls.
How many participants, which segments, and how were they selected? Example: '500 randomly sampled active users in the US mobile app.'
How are participants informed or consenting? Choose the closest match and describe specifics below.
Paste or summarize the consent text, opt-in flow, or notice language. Include how to withdraw and contact details.
List the concrete signals, metrics and logs you'll monitor (with units). Example: 'payment error rate (%), login failures per minute, API 5xx rate, customer support complaint volume.'
How often will the signals be checked or automated alerts evaluated?
State numeric thresholds for primary signals and what each level means. Example: 'Payment error rate > 0.5% = warning; > 1.0% = critical (rollback).'
Actions or safeguards to attempt before rolling back (feature flags, throttling, partial disabling, retries, rate limits). Example: 'Disable new recommendation model via feature flag; revert config to previous model.'
List the exact conditions that will cause an immediate rollback. Use the thresholds you entered above. Example: 'Critical payment error threshold exceeded for 15 consecutive minutes OR >5% of active sessions experiencing server errors.'
Provide numbered, testable steps including who executes them, how to confirm rollback completed, and where to log the action. Example: '1) Turn off feature flag X (owner: SRE). 2) Revert service config to commit abc123 (owner: backend lead). 3) Validate error rate returns to baseline within 10 minutes.'
Who is on call for monitoring, who can authorize rollback, legal and comms contacts, and escalation path (include phone/email).
What stakeholders will be notified, what will be said, and who approves externally facing messages (if any). Include timing (e.g., 'Notify exec team within 30 minutes of critical trigger').
Confirm whether legal/compliance has reviewed the pilot for regulatory or contractual exposure.
Summarize any conditions, required notices, or restrictions from legal/compliance. If not applicable, explain why.
How will you evaluate results, measure learning, and decide next steps? Include specific metrics and minimum sample sizes if relevant.
Under what conditions will the pilot end normally (timebox, sample size, metric targets)?
Person who authorizes this pilot and is empowered to approve rollback decisions.
YYYY-MM-DD
Links to runbooks, dashboards, consent forms, test plans, or incident tickets. Store URLs or internal references here.
You can explore this tool now. Sign in or create an account to save your responses and return to them later.
Sign In to Save Create an Account
Make this tool part of your work

Save a personal copy, bring it to your team, or tailor the questions and workflow to fit what you are hungry to improve.

Member customization and team collaboration are coming soon.
Discuss Ask the Librarian

Discussion

Comments and conversation will live here.