What Is Sporadic Demand in FBA Inventory?
Sporadic demand describes a sales pattern where orders arrive infrequently, separated by stretches of zero activity, but the order size is relatively consistent each time a sale does occur. If a SKU sells 4-6 units every 10 to 18 days with nothing in between, that is sporadic demand.
Sporadic demand sits in a specific quadrant of the Syntetos-Boylan demand classification framework. You identify it by calculating two metrics from your sales data:
- ADI (Average Demand Interval): The average number of periods between non-zero demand occurrences. Sporadic items have ADI > 1.32.
- CV² of demand sizes: The squared coefficient of variation of the non-zero demand quantities. Sporadic items have CV² < 0.49 (meaning when they sell, the quantity is fairly consistent).
This distinction matters because sporadic demand requires different forecasting and safety stock calculations than smooth, erratic, or lumpy demand patterns. Using a standard moving average on sporadic data understates true demand by averaging in all those zero periods.
How to Identify Sporadic Demand from Sales Data
The two formulas you need:
ADI = Total number of periods / Number of periods with non-zero demand
If you have 26 weeks of data and sales occurred in 12 of those weeks, ADI = 26/12 = 2.17. The threshold is 1.32: anything above means demand is intermittent.
CV² = (Standard Deviation of non-zero demand / Mean of non-zero demand)²
Only include the weeks where actual sales occurred. If those 12 selling weeks had quantities of 4, 5, 6, 4, 5, 5, 4, 6, 5, 4, 5, 5, the mean is 4.83 and std dev is 0.72. CV² = (0.72/4.83)² = 0.022. Well below 0.49, confirming consistent demand sizes.
| Pattern | ADI | CV² | Best Forecast Method |
|---|---|---|---|
| Smooth | ≤ 1.32 | < 0.49 | Exponential smoothing, moving average |
| Erratic | ≤ 1.32 | ≥ 0.49 | Weighted moving average, outlier-adjusted |
| Sporadic (intermittent) | > 1.32 | < 0.49 | Croston's method |
| Lumpy | > 1.32 | ≥ 0.49 | Croston's SBA variant |
Worked Example: Forecasting Sporadic Demand for an FBA SKU
A seller stocks a $45 specialty kitchen tool with a 60-day lead time from Guangzhou. Here are 26 weeks of unit sales:
0, 0, 5, 0, 0, 4, 0, 0, 0, 5, 0, 0, 6, 0, 0, 0, 4, 0, 5, 0, 0, 0, 5, 0, 0, 4
Step 1: Calculate ADI. Non-zero weeks: 8 out of 26. ADI = 26/8 = 3.25. Above 1.32, so demand is intermittent.
Step 2: Calculate CV². Non-zero quantities: 5, 4, 5, 6, 4, 5, 5, 4. Mean = 4.75, Std Dev = 0.71. CV² = (0.71/4.75)² = 0.022. Below 0.49, so demand sizes are consistent. Classification: sporadic.
Step 3: Apply Croston's method. Separate into two forecasts: demand size (avg ~4.75 units per occurrence) and inter-arrival interval (avg ~3.25 weeks between sales). Expected weekly demand = 4.75 / 3.25 = 1.46 units/week.
Step 4: Calculate lead time demand. 60-day lead time = ~8.6 weeks. Lead time demand = 1.46 x 8.6 = 12.6 units. Add safety stock for the intermittent pattern (use a Poisson-based method rather than normal distribution): approximately 8 additional units at a 95% service level.
Reorder point: 21 units. Reorder quantity: 38 units (covers ~26 weeks).
A simple moving average would have calculated weekly demand as (total units / 26 weeks) = 38/26 = 1.46, which happens to match Croston's here. But the safety stock calculation differs significantly because Croston's accounts for the intermittent arrival pattern, producing a more appropriate buffer.
Why Sporadic Demand Is Common in FBA
Sporadic demand shows up frequently in FBA catalogs because of the marketplace's long-tail dynamics. Niche products, specialty tools, and higher-ASP items naturally attract fewer but more consistent buyers. A $67 knife sharpener does not sell like a $12 phone case, but it may generate more profit per sale.
The danger with sporadic items in FBA is that Amazon's own restock recommendations treat every SKU the same way: a simple velocity calculation that divides total units by total days. This averages in all the zero-sales periods, producing an artificially low demand estimate. Sellers who blindly follow Amazon's suggestion end up chronically understocked on sporadic products.
Conversely, the sell-through rate for sporadic items will always look low because inventory sits between sales events. This drags down the IPI score. The solution is not to avoid these products but to size your FBA inventory correctly: hold just enough to cover lead time demand plus an intermittent-aware safety buffer, and consider hybrid fulfillment (FBA for a small buffer, FBM or 3PL for backup stock).
Common Sporadic Demand Mistakes
1. Using a moving average on sporadic data. A 4-week moving average for the SKU above would read 0, 0, 0, 4 in the most recent window = 1.0 units/week. But the actual expected demand rate is 1.46 units/week. The moving average is depressed by the zeros, leading to chronic understocking and missed sales.
2. Confusing sporadic with declining demand. A string of zero-sales weeks looks alarming in a dashboard. Some sellers interpret this as "this product is dying" and stop reordering. But sporadic items have always looked like this. Before cutting a SKU, check whether the inter-arrival interval has actually lengthened compared to the prior 6 months, or if this is just the normal pattern.
3. Overstocking after a cluster of orders. Sporadic items sometimes produce back-to-back sales events by coincidence. Two sales in one week does not mean demand has doubled. Croston's method smooths this naturally by updating both the demand size and interval estimates. Reacting to a cluster by doubling your next order creates excess inventory that sits through the next dry spell.
Related Glossary Terms
Frequently Asked Questions
How do I tell if a SKU has sporadic demand?
Calculate ADI (Average Demand Interval) and CV² of non-zero demand sizes from your weekly sales data. If ADI is above 1.32 and CV² is below 0.49, the SKU has sporadic demand. In plain terms: it sells infrequently but in consistent quantities when it does sell.
What forecasting method works for sporadic demand?
Croston's method is the standard approach for sporadic demand. It separates the forecast into two components: the expected demand size (when a sale occurs) and the expected interval between sales. This avoids the problem of averaging in all the zero-sales weeks that ruins standard moving average forecasts.
Should I keep sporadic demand SKUs in FBA?
It depends on margin per sale versus holding cost. A sporadic SKU selling 5 units every 3 weeks at $65 with 35% margin generates real profit. But if storage fees and aged surcharges consume that margin, transition it to FBM or a 3PL where holding costs are lower.
How is sporadic demand different from lumpy demand?
Both are intermittent (many zero-sales periods), but sporadic demand has consistent order sizes when sales occur, while lumpy demand has highly variable order sizes. A sporadic SKU might sell 5 units every 2-3 weeks. A lumpy SKU might sell 3 units one time and 40 the next.
Why does Amazon's restock tool fail for sporadic items?
Amazon's recommended restock quantity uses a simple velocity calculation that averages in zero-sales days. For a sporadic SKU, this produces an artificially low demand estimate that leads to understocking. Specialized intermittent demand forecasting like Croston's method is required.