How do I forecast Amazon demand accurately?
The most accurate simple forecast weights the last 30 days at 60% and the prior 60 days at 40%, then adjusts for seasonality and known promotions. For most private-label SKUs this beats gut-feel and matches what statistical models produce within a 10% margin. Rebuild the forecast every 2 to 4 weeks.
The simple weighted formula
Forecast = (last 30-day velocity x 0.6) + (60-day prior velocity x 0.4). Multiply by seasonality index if you have one (e.g. 1.4 for Q4 on gift SKUs). Add units for known promotions.
What to layer on top
Lead time buffer: multiply forecast by lead-time days plus a 15% safety margin. Stockout adjustment: exclude out-of-stock days from the velocity denominator. Return adjustment: subtract expected refund rate from forecasted units.
When to use a statistical model instead
SKUs with heavy seasonality (Halloween, Christmas), SKUs launching soon, and SKUs with wide day-to-day variance benefit from something more advanced. For mature stable SKUs, the weighted formula performs within 10% of statistical models.
Frequently asked questions
How far ahead should I forecast?
Cover lead time + 30 days as the minimum. For long-lead SKUs from Asia, forecast 120 days out.
How often should I rebuild the forecast?
Every 2 to 4 weeks for stable SKUs, weekly during launches or seasonality peaks.
What is the biggest forecasting mistake?
Extrapolating a Q4 spike into Q1 planning. Always deseasonalise before projecting forward.
Does Nova generate forecasts?
Nova exposes velocity, sell-through, and days of cover so you can build forecasts. For automated statistical forecasts, use dedicated forecasting tools alongside.
See this in Nova
Nova surfaces this metric per SKU in near real time so you can act before your next monthly close catches it.
Try Nova free for 14 daysRelated questions
What is SKU velocity and why does it matter?
SKU velocity is the average units sold per day for a specific SKU, usually measured over a rolling 30-day window. It drives every downstream inventory decision: PO size, safety stock, days of cover targets, and IPI score. A SKU whose velocity drops 30% for two weeks is either losing search rank or facing seasonality.
What is days of cover and how do I calculate it?
Days of cover = current on-hand inventory / average daily units sold. If you have 600 units on hand and sell 20/day, you have 30 days of cover. Most sellers target 45 to 75 days of cover for FBA SKUs: enough to survive a demand spike or a shipping delay, not so much that IPI or storage fees suffer.
What is a good sell-through rate on Amazon?
Amazon's sell-through rate is units shipped in the last 90 days divided by average inventory over the same period. Above 3.0 is strong (inventory turns 12x/year). 2.0 to 3.0 is healthy. Below 1.0 usually means overordered units and forecasts long-term storage fees within a quarter.
How do I avoid FBA storage limits?
FBA storage limits are set by IPI score. Keep IPI above 400 by clearing stranded inventory weekly, running removals or liquidations on units aged 180+ days, and keeping your top 20 SKUs consistently in stock. Restock-limited accounts often gain 30% to 60% more capacity within 30 days after a targeted cleanup.
What are Amazon FBA restock limits?
Restock limits cap how many units you can send to FBA per storage type (standard, oversize, apparel, footwear). Amazon sets them based on IPI, sell-through, and available warehouse capacity. Limits refresh weekly. Selling faster and keeping IPI above 400 is the only reliable way to raise them.