Kanban sizing calculator & risk checklist

Purpose

Use this calculator to convert demand and lead-time inputs into defensible kanban quantities and card counts, explicitly accounting for variability and your target service level. The page also includes practical rules, a worked example, a spreadsheet template layout, and a concise risk checklist for safe pilots.

Key variables (define before you calculate)

• D — average demand per period (units per day, week, or whatever period you use). Be consistent with lead time units.
• L — average supplier lead time in the same periods as D (e.g., days).
• ContainerQty — quantity per kanban container or order batch (units per card).
• Service level (SL) — desired probability of not stocking out during lead time (e.g., 95%).
• σ_d — standard deviation of demand per period. If unknown, estimate from historical demand or use a coefficient of variation (CV = σ_d / D).
• σ_L — standard deviation of lead time in periods (supplier variability). If you only have min/max lead times, approximate σ_L ≈ (maxL - minL)/4 as a quick heuristic.

Translate service level to Z (standard normal z-score)

Common conversions: 90% → 1.28, 95% → 1.65, 97.5% → 1.96, 99% → 2.33. Use an online z-table if you need other values.

Core formulas

These formulas compute the number of kanban cards required when you account for demand and lead-time variability.

1. Expected demand during lead time: μ_DL = D × L
2. Variance of demand during lead time (accounts for demand and lead-time variability):

   σ_DL^2 = L × σ_d^2 + D^2 × σ_L^2

   So σ_DL = sqrt(L × σ_d^2 + D^2 × σ_L^2).

3. Safety stock for your service level: SS = Z × σ_DL.
4. Total inventory to cover lead time and safety: Total = μ_DL + SS.
5. Number of kanban cards (round up to whole cards):

   Cards = ceil( Total / ContainerQty )

Worked example

Assume:

• D = 100 units/day
• L = 3 days
• ContainerQty = 50 units per kanban container
• σ_d = 20 units/day (demand std dev)
• σ_L = 1 day (lead time std dev)
• Service level = 95% → Z = 1.65

Step calculations:

1. μ_DL = 100 × 3 = 300 units
2. σ_DL^2 = 3 × (20^2) + (100^2) × (1^2) = 3 × 400 + 10,000 = 1,200 + 10,000 = 11,200
   σ_DL = sqrt(11,200) ≈ 105.8 units
3. SS = 1.65 × 105.8 ≈ 174.6 → round to 175 units
4. Total = 300 + 175 = 475 units
5. Cards = ceil(475 / 50) = ceil(9.5) = 10 cards

Result: Start with 10 kanban cards (10 containers of 50 units) for this SKU under the given assumptions.

When you don't have full variance data (quick heuristic)

If you only know average demand and an approximate sense of risk, use a buffer multiplier:

Cards ≈ ceil( D × L × (1 + buffer) / ContainerQty )

• Low variability suppliers: buffer 5–10%
• Moderate variability: buffer 10–25%
• High variability / long lead time / critical SKU: buffer 25–50% (use cautiously)

Heuristics are temporary. Replace with statistical inputs as you collect data during the pilot.

Multi-SKU and multi-supplier notes

• Calculate cards per SKU-Supplier combination (each kanban card typically represents a single SKU packaged for a single supplier).