BFCM readiness checklist

Copy this checklist into your team’s project management tool, assign owners, and work through it in order. Every item links to the Academy page that explains how to complete it in MaxoPerf.

T-6 weeks: Foundation

Traffic model complete. Peak VU count and RPS targets derived from last year’s data and growth forecast. See Capacity planning and traffic modeling.
Traffic model documented. Model document stored in team wiki with source data, growth factor, safety headroom, and final VU/RPS numbers. All engineers using the same numbers.
Target environment confirmed. Staging environment is production-like in configuration, data volume, and downstream connectivity. Database seeded with representative data volume.
Third-party authorizations obtained. Written authorization from any third-party APIs (payment processor, tax, fraud) that will be hit with test traffic. See Third-party and payment dependencies.
Mock servers deployed. Payment gateway, tax, shipping, and fraud mocks deployed in staging with realistic response times and error rates.
Baseline load test run. Full-funnel load test at 1× modeled peak VU count passes. Run tagged as bfcm-baseline-2025 in MaxoPerf. See End-to-end journey load.
Failure criteria configured. Per-label failure criteria set on all BFCM tests — at minimum p95 latency and error rate thresholds. See Failure criteria pass/fail gates.

T-4 weeks: Load validation

Load profiles configured. Doorbuster spike profile, email wave profile, and sustained surge profile are created as MaxoPerf tests. See Realistic peak load profiles.
Load test at modeled peak passes. Full-funnel load test at 1× peak VU count: failure criteria green, p95 within SLO, error rate < 0.5 %. This is the year-over-year comparison baseline.

Smoke test for every new BFCM feature. Any feature that was released since last year’s BFCM — new checkout flow, new product recommendation panel, new loyalty integration — has been smoke-tested and passes.
CDN cache warm-up plan documented. Pre-warming strategy for homepage, category pages, and static assets is written and tested. Cache hit rate confirmed > 90 % for these assets under load.

T-3 weeks: Stress and spike

Stress test run. Gradual ramp to 200 % of modeled peak. Breaking-point VU count documented. Headroom confirmed > 1.5× above modeled peak. See Spike and stress for sales.
Spike test run (doorbuster profile). Near-instantaneous ramp to 5× modeled peak. Recovery observed and documented. See Spike test.
Bottlenecks identified. Any endpoint failing at < 1.5× modeled peak is documented with root cause analysis.
Fixes deployed. All identified bottlenecks fixed, tested, and confirmed not to regress existing behavior.
Re-test after fixes. Stress and spike tests re-run after bottleneck fixes. All passing. Go/No-Go gate: must be green before proceeding to T-1 week.

T-1 week: Dress rehearsal

Soak test complete. 8+ hour soak at modeled peak VU count. No drift > 10 % in p95 latency. No errors in final 2 hours. See Soak and stability.
Dress rehearsal run. Multi-wave profile (doorbuster + email wave + sustained) run for 3+ hours. All failure criteria green. See Realistic peak load profiles.
War-room runbook complete. Monitoring assignment matrix, abort criteria, and action procedures written, reviewed, and signed off by engineering lead and business lead. See War-room and runbook.
War-room rehearsed. Dress rehearsal included a full war-room exercise with amber condition injection. Every engineer practiced their role.
On-call schedule confirmed. Black Friday and Cyber Monday on-call schedule populated with names, phone numbers, and escalation order.
Rollback procedures tested. Emergency rollback for the most recent release is tested and known to work.
Year-over-year comparison baseline pinned. The T-1 week dress rehearsal run (or the T-4 week baseline) is pinned in MaxoPerf as the comparison baseline for game day.

Go/No-Go gate: all T-1 week items must be complete. Any red item requires explicit decision from engineering lead and business lead before proceeding.

Day-of event

War-room active. All engineers in monitoring seats by 30 minutes before sale start.
Monitoring dashboards open. MaxoPerf active-runs view, infrastructure metrics, and APM dashboards all open and refreshed.
Pre-sale smoke test run. A lightweight smoke test (1–5 VUs) run 60 minutes before sale start to confirm every endpoint is responding. Result: Finished, no errors.
CDN cache pre-warmed. Cache warming script executed. Confirmed cache hit rate > 90 % on sale landing pages.
Alert thresholds armed. All alerting rules confirmed active. Test alert fired and received by every engineer.
Runbook accessible offline. Runbook printed or available on a device not dependent on systems under load.

During the event

5-minute metric check rotation active. Each engineer checking their assigned metric every 5 minutes and posting status.
Any amber event logged. Start time, metric, threshold, action taken, and resolution time.
Any red event handled per runbook. Runbook procedure executed without deviation.
MaxoPerf runs active. Synthetic monitoring tests running throughout the event.

Post-event (T+1 week)

Year-over-year comparison run. This year’s peak game-day run compared to last year’s baseline in MaxoPerf. Delta chips reviewed and documented.

Incident log complete. All amber and red events documented with root cause and resolution.
Runbook updated. Any actions taken during the event that were not in the runbook are added.
Game-day run archived. Tagged bfcm-2025-game-day in MaxoPerf for next year’s year-over-year comparison.
Post-mortem scheduled. Blameless post-mortem meeting scheduled within 5 business days.
T+1 year reminder set. Calendar reminder set for T-8 weeks before next year’s BFCM to start the readiness program.

Where to go next

Daily countdown calendar — the week-by-week schedule of test activities.
War-room and runbook — the full war-room setup guide.
Peak events do and don’t — common mistakes to avoid across the program.
Comparing runs and baselines — how to run the year-over-year comparison.