Amazon gives vendors a significant amount of information about expected customer demand. Those forecasts can inform production and inventory planning, but they don’t tell a vendor how much inventory to carry.
To make that decision well, vendors need to understand what Amazon’s forecasts represent, what they don’t, and the economic consequences of carrying too much or too little inventory.
This Guide explains how to turn Amazon’s forecast into better inventory decisions.
The Forecast report in Vendor Central estimates future customer demand for an ASIN.
It’s intended to help vendors make production and inventory planning decisions. It is not a forecast of how much Amazon will order.
Amazon describes the forecast as unconstrained customer demand. In simple terms, it estimates what customers may want to buy, independent of how much inventory Amazon already has or has on order. Amazon considers those and other factors separately when determining how much inventory it needs to purchase.
That’s why the quantities shown in the forecast may differ significantly from subsequent purchase orders. The forecast is an estimate of customer demand, not a promise of future orders.
Knowing what Amazon forecasts is only the beginning. Vendors must also understand how much uncertainty the forecast represents and how relevant that information is to a decision being made today.
That requires understanding three things: the probability represented by the forecast, the time horizon being considered, and how Amazon's forecasts have historically performed for the vendor's products.
Rather than providing a single prediction, Amazon presents several views of possible demand:
If the P70 forecast for an ASIN were 100 units, for example, there would be approximately a 70% probability that demand will be at or below 100 units, and a 30% probability that demand will exceed 100 units. Moving from P70 to P80 or P90 raises the threshold so that a larger share of modeled demand outcomes falls at or below it.
Higher P-values therefore provide greater protection against demand exceeding the planned quantity. They do not represent increasingly accurate predictions, nor do they mean Amazon expects the higher quantity to sell.
Forecast availability is not forecast reliability.
A forecast may extend far enough into the future to support annual planning, production scheduling, or raw-material decisions. That does not mean every period in the forecast should carry equal decision weight.
Uncertainty generally increases as the decision moves farther into the future. More time creates more opportunity for customer demand, pricing, competition, promotions, seasonality, and other conditions to depart from current assumptions.
Yet waiting for better information is not always possible. A business with long manufacturing or transportation lead times may need to commit capacity months before demand occurs. By the time the forecast becomes more reliable, the organization may no longer be able to respond.
The practical question is not how many weeks of forecast Amazon provides. It’s: What decision must we make today, and how far into the future does that decision require us to look?
A vendor allocating available finished goods for next week can wait for information that a vendor planning a six-month production run cannot. The farther in advance the commitment must be made, the more uncertainty the organization may need to accept, and the more valuable flexibility becomes.
Forecasts will be wrong. But error alone doesn’t make a forecast useless.
How Amazon’s forecasts perform for the vendor’s own products should improve how the organization interprets them over time. Where does Amazon consistently forecast high or low? By how much? At what horizon? Does performance change around particular ASINs, launches, promotions, or seasonal periods?
The objective is not to build a forecasting-science function around Amazon’s data. It’s to use accumulated evidence to give the forecast the appropriate weight in future inventory decisions.
P-values help describe possible demand outcomes. Time affects how much weight the forecast should carry for a decision being made today. The organization’s own experience can further refine how it interprets both.
But none of those three dimensions determines how much inventory the vendor should carry.
That requires a commercial decision.
Understanding Amazon’s forecast establishes what future demand may look like. The next step is deciding how much inventory the organization is willing to commit against that opportunity and uncertainty.
Vendors often collapse three different decisions into one:
Forecast → Inventory Plan → Supply Commitment
The forecast provides information about possible customer demand. The inventory plan defines the inventory position the vendor wants against that opportunity and uncertainty. The supply commitment is what the organization ultimately produces, reserves, allocates, or makes available.
Those quantities are related, but they’re not interchangeable.
The inventory plan sits between information and commitment. Creating it requires leadership to answer four questions:
Those questions should generally be answered at the ASIN level. An account-wide rule such as “we plan Amazon to P80” may be administratively convenient, but it ignores economically important differences between products.
Ask: What are the economic consequences if we are wrong in either direction?
The cost of being wrong is not symmetrical. Carrying too little inventory creates one set of economic consequences. Carrying too much creates another.
Too Little Inventory
The immediate cost of insufficient inventory is the contribution profit lost on demand the business cannot serve. The total economic loss may be greater.
Customers may substitute a competitor’s product, delay the purchase, buy from another retailer, or abandon the transaction. Inventory arriving later may capture future sales, but it can’t guarantee recovery of the demand that existed while the product was unavailable.
A stockout can also weaken the return on investments already made to create or capture demand. Advertising, promotions, content, and launch activity have less commercial value when the organization lacks the inventory required to serve the opportunity they produce.
If a vendor invests to create or capture demand without ensuring the inventory required to serve it, its commercial investments can begin working against one another.
Too Much Inventory
Excess inventory creates costs even if it eventually sells.
It commits working capital, production capacity, warehouse space, and available units that cannot be used elsewhere. It may also create carrying costs, markdown exposure, obsolescence, or shelf-life risk.
The relevant question is not merely whether the inventory will sell eventually. It’s whether the return ultimately generated justified the cost and risk of carrying it.
The inventory decision therefore requires weighing the economic cost of being wrong in both directions. Too little inventory risks leaving valuable demand unserved. Too much creates costs and commitments that may not be justified by the demand ultimately realized.
Ask: How quickly and economically can the organization respond when actual demand differs from the plan?
Forecast uncertainty describes how widely demand may vary. Replenishment flexibility determines how long the organization must remain exposed when the inventory plan proves wrong.
A vendor with available finished goods, short transportation lead times, flexible suppliers, and the ability to redirect inventory among channels may carry less protection against upside demand. If sales accelerate, it can replenish.
Another vendor may need to commit raw materials, production capacity, and transportation months in advance. Minimum production quantities or specialized manufacturing may make the decision difficult to reverse. For that business, the initial inventory position carries greater consequences because there are fewer opportunities to correct it later.
Flexibility may come from shorter production and transportation lead times, dependable suppliers, available raw materials or finished goods, smaller production minimums, transferable inventory, or longer shelf life.
Greater flexibility reduces how much uncertainty the organization may need to protect against in advance. Less flexibility increases the importance of making the initial commitment correctly.
Flexibility has economic value.
Ask: How important is it that this particular ASIN remain available?
Products play different economic and strategic roles within an Amazon portfolio. Contribution economics matter, but so do hero-product status, repeat-purchase behavior, launch priorities, lifecycle, and the investments the organization is making to create or capture demand.
Consider two ASINs with identical forecast distributions.
ASIN A generates strong contribution profit, plays an important strategic role, supports repeat purchasing, and is receiving meaningful commercial investment. Running out of stock would undermine an important source of economic value and customer demand.
ASIN B generates weaker contribution, has limited strategic importance, and plays a marginal role in the portfolio. Maintaining the same level of availability carries less commercial importance.
The forecasts are the same.
The rational inventory decisions are not.
The appropriate position depends not only on what demand may occur, but on how much the organization values maintaining availability for that particular product.
Ask: What else could the organization do with the resources required to support this inventory position?
Inventory is a commercial investment. Every dollar of working capital invested in inventory is capital that cannot simultaneously fund advertising, promotions, content, or another commercial priority. Every unit of available inventory or production capacity committed to one ASIN or channel may also be unavailable to another.
The question is not simply whether Amazon might eventually sell the inventory. It’s whether the expected return justifies committing those resources to this opportunity instead of another.
That leads to a more useful question: Is this the best use of the next unit of inventory, capacity, or capital?
When inventory and capacity are plentiful, the opportunity cost of additional protection may be modest. As those resources become constrained, inventory planning increasingly becomes an allocation decision.
Across channels, leadership must decide how available inventory should be divided among Amazon, direct-to-consumer, distributors, retail partners, and other marketplaces. Within Amazon, it must decide which ASINs deserve the greatest protection.
Strategy should influence which shortages the organization is willing to tolerate.
If leadership describes Amazon as a strategic growth channel while the operating process consistently gives it only the inventory remaining after every other channel is served, resource allocation contradicts stated strategy. The same problem exists within Amazon when hero products, launches, and marginal ASINs receive identical protection despite materially different economics and strategic roles.
Amazon should not receive priority by default. Neither should any other channel or product. Inventory allocation should reflect the organization’s strategy and the economic value of the opportunities competing for supply.
Before committing inventory against an ASIN, leadership should be able to answer six questions:
The purpose of these questions is not to create a more elaborate forecast. It’s to convert information about uncertain demand into a deliberate commercial decision.
Vendors often look to Amazon’s forecast for the number they should use. No forecast can supply that answer on its own.
Amazon can estimate the range of possible customer demand. Vendor leadership must decide how valuable that opportunity is, what being wrong would cost, how readily the organization can respond, and what resources the opportunity deserves.
The forecast informs the decision.
Vendors must make the decision.
