Amazon FBA Inventory Forecasting: How to Avoid Stockouts and Overstock
The Real Cost of Stockouts
Running out of stock on Amazon is not just a missed sale. It triggers a cascade of consequences that can take weeks or months to recover from.
When your listing goes to zero inventory, Amazon suppresses it in search results. Your Best Seller Rank (BSR) plummets. Competitors capture the customers who would have bought from you, and some of those customers become repeat buyers for your competitors. Your PPC campaigns pause automatically because there is nothing to sell, and when you restock, you have to rebuild momentum from a lower baseline.
The data is stark. An analysis of seller accounts shows that a seven-day stockout on a product selling 20 units per day typically requires 14 to 21 days of recovery to return to pre-stockout sales velocity. During that recovery period, you are selling fewer units, ranking lower, and often spending more on advertising to climb back.
For a product with $15 profit per unit, a single week-long stockout can cost over $4,000 in lost profit between the stockout itself and the recovery period.
The Equally Expensive Problem of Overstock
If stockouts are the acute pain, overstock is the chronic disease. Having too much inventory at Amazon feels safe, but the costs compound silently.
Monthly storage fees at $0.87 per cubic foot during off-peak and $2.40 during Q4 do not seem alarming on a per-unit basis. But multiply that across hundreds of excess units sitting for months, and you are looking at thousands of dollars in avoidable storage costs. Once inventory crosses the 181-day threshold, the aged inventory surcharge kicks in, and your per-unit cost increases dramatically.
Beyond fees, overstocked capital is capital you cannot use for better-performing products. Every dollar tied up in slow-moving FBA inventory is a dollar not available for sourcing the products that actually drive your business forward.
Core Forecasting Concepts
Velocity-Based Demand Estimation
The foundation of any forecast is your sales velocity: how many units of each product you sell per day. The simplest approach is taking the average daily sales over the last 30 days. This works for stable products but fails badly for anything with trends or seasonality.
A better approach uses weighted averages that emphasize recent data. If your last four weeks of daily sales were 8, 10, 12, and 15 units per day, a simple average says 11.25. A weighted average that gives 40% weight to the most recent week, 30% to the week before, 20% to the week before that, and 10% to the earliest week gives 12.3, which better reflects the upward trend.
Trend Detection
Is your product's demand increasing, decreasing, or flat? The answer changes your reorder quantity significantly.
Upward trend: Increase your forecast by the rate of growth. If sales have been growing 10% week-over-week, your four-week forecast should reflect continued growth, not just the current average.
Downward trend: Reduce your reorder to avoid building excess inventory. A declining product needs smaller, more frequent orders rather than a single large replenishment.
Flat/stable: Your standard forecast based on recent averages is likely accurate. Focus on getting safety stock right.
Seasonality Adjustments
Products with seasonal demand patterns need forecasts that look beyond the recent trend. A product that sells 20 units per day in February and 50 units per day in November needs a restock plan that accounts for the upcoming demand spike, not just current velocity.
Build a seasonal index by comparing each month's historical sales to the annual average. If March historically represents 8% of annual sales and June represents 12%, your June forecast should be 50% higher than March even if current sales are flat.
Lead Time Accuracy
Your lead time is the total time from placing a restock order to having inventory available for sale on Amazon. This includes:
- Supplier production time: The time to manufacture or pull your order
- Shipping time: Transit from supplier to your prep center or Amazon
- Prep time: Labeling, inspection, and packaging for FBA
- Amazon receiving time: The time from delivery to available inventory (typically 5 to 14 days, but can spike to 21+ days during peak periods)
Most sellers underestimate their lead time by ignoring Amazon's receiving delay. If your supplier delivers in 14 days and you assume Amazon receives in 3 days, but the actual average is 10 days, you are planning on a 17-day lead time when reality is 24 days. That seven-day gap is exactly how stockouts happen.
Track your actual lead times for each supplier and each Amazon fulfillment center. The average is not enough — you need to know the variability.
Safety Stock: Your Stockout Insurance
Safety stock is the extra inventory buffer you maintain above your expected demand during lead time. It exists to protect against two types of uncertainty: demand variability (sales might be higher than forecast) and supply variability (lead time might be longer than expected).
The Basic Safety Stock Formula
Safety Stock = Z x sigma_d x sqrt(L)
Where:
- Z = service level factor (1.65 for 95% in-stock rate, 2.33 for 99%)
- sigma_d = standard deviation of daily demand
- L = lead time in days
For a product with daily demand averaging 15 units (standard deviation of 4), a 21-day lead time, and a target 95% in-stock rate:
Safety Stock = 1.65 x 4 x sqrt(21) = 1.65 x 4 x 4.58 = 30 units
Service Level Tradeoffs
A 99% service level sounds ideal, but it requires significantly more safety stock than 95%. The right target depends on the product:
| Service Level | Z-Score | Relative Inventory |
|---|---|---|
| 90% | 1.28 | Baseline |
| 95% | 1.65 | 29% more than 90% |
| 98% | 2.05 | 60% more than 90% |
| 99% | 2.33 | 82% more than 90% |
For your top 20% of products (by profit contribution), target 98-99%. For the long tail, 90-95% is often sufficient. The capital savings from lower safety stock on C-class products can be redirected to ensuring your A-class products never stock out.
ABC-XYZ Classification for Inventory Prioritization
Not all products deserve the same inventory management attention. ABC-XYZ classification segments your catalog on two dimensions:
ABC (Revenue/Profit Contribution):
- A items: Top 20% of products that generate 80% of revenue or profit
- B items: Middle 30% contributing 15% of revenue
- C items: Bottom 50% contributing 5% of revenue
XYZ (Demand Variability):
- X items: Stable, predictable demand (coefficient of variation below 0.5)
- Y items: Moderate variability (CV between 0.5 and 1.0)
- Z items: Highly erratic or sporadic demand (CV above 1.0)
The intersection creates nine segments, each requiring a different inventory strategy:
| Segment | Strategy |
|---|---|
| AX | High stock, automated reordering, tight monitoring |
| AY | Moderate stock, frequent review, trend monitoring |
| AZ | Conservative stock, manual review, consider dropping |
| BX | Moderate stock, periodic review |
| BY | Standard stock, regular review |
| BZ | Low stock, evaluate if worth carrying |
| CX | Low stock, automated with min/max levels |
| CY | Minimal stock, order as needed |
| CZ | Consider discontinuing or remove from FBA |
Monte Carlo Simulation for Demand Uncertainty
For critical products where the cost of a stockout is high, simple formulas may not capture the full range of possible outcomes. Monte Carlo simulation runs thousands of scenarios with randomized demand and lead time values to give you a probability distribution of potential outcomes.
Instead of a single number like "you need 450 units for the next 30 days," Monte Carlo tells you there is a 50% chance you need between 400 and 500 units, a 90% chance you need between 350 and 550, and a 99% chance you need between 300 and 620.
This lets you make informed decisions based on your risk tolerance rather than a single point estimate.
How SellerVault Handles Forecasting
SellerVault calculates restock recommendations using velocity-based demand forecasting with trend and seasonality adjustments. It tracks your actual lead times per supplier, automatically classifies products using ABC-XYZ segmentation, and runs Monte Carlo simulations for high-priority SKUs.
The restock dashboard shows you exactly what to order, when, and how much, with urgency scoring that highlights products at risk of stockout. You can view recommendations in list, card, calendar, or kanban views depending on your workflow.
Ready to stop guessing on inventory? Start your free trial and let SellerVault's forecasting engine manage your restock planning, or compare plans to get started.