Write a Load Test Plan for Your Checkout Before Black Friday Traffic Hits

You are a senior performance engineer who helps small ecommerce teams get their checkout ready for holiday traffic peaks. I want a practical load test plan for our checkout that I can run in the weeks before Black Friday, using only the details I give you. Do not invent traffic numbers, infrastructure limits, or vendor quotas. If something is missing, mark it [NEEDS INPUT] and tell me how to measure it.

Stack and hosting (framework, database, cache, CDN, hosting provider): [STACK]
Checkout flow steps from cart to confirmation, including third party calls (payments, tax, shipping rates, fraud checks): [FLOW]
Last year's peak traffic or best estimate, with where the number comes from: [PEAK]
Expected multiplier for this year and why: [GROWTH]
Load testing tool we prefer or can use (k6, Locust, JMeter, Artillery, none yet): [TOOL]
Environments available (staging that mirrors production, production with a test mode, neither): [ENVIRONMENTS]
Rules from our payment provider about testing, if I know them: [PAYMENT RULES]

Requirements:
1. Turn [PEAK] and [GROWTH] into target numbers: concurrent users, requests per second for each checkout step, and the p95 and p99 response times we should hold. Show the math.
2. Define 4 test types with clear pass or fail criteria: a baseline test, a ramp up load test, a spike test that mimics a sale going live, and a one hour soak test.
3. Write a starter script in [TOOL] (or k6 if none) that walks the checkout steps with realistic think time, test accounts, and test products. Never use real customer data or real cards. Use the payment provider's sandbox or test card numbers, and stub any third party that forbids load testing.
4. List the metrics to watch during each run: error rate by step, database connections and slow queries, cache hit rate, queue depth, CPU and memory, and third party latency.
5. Explain how to find the first bottleneck from those metrics and give the 5 most common fixes for checkout slowdowns (missing indexes, N plus 1 queries, unbounded connection pools, synchronous third party calls, uncached shipping or tax lookups).
6. Add a safety section: run outside business hours, warn the hosting provider if their terms require it, set a kill switch, and never test against production payments.
7. Create a go live checklist for the week of the sale: feature freeze, scaling settings confirmed, alerts set, on call rota, and a rollback plan.

Deliverables:
1. Target numbers with calculations
2. Test matrix with pass or fail criteria
3. Starter load test script
4. Monitoring and bottleneck guide
5. Safety rules and go live checklist