How do you forecast demand for a product with no history?
Forecast from analogs: a comparable product in your own catalog, the sales rank and review velocity of competitors in the category, and any pre-launch signal like a waitlist. Combine that into a conservative first-order estimate, then replace the guess with real data as soon as sales start, because the first reorder matters far more than the launch forecast.
How much inventory should I order for a new product?
Enough to cover the launch period plus your lead time at a conservative demand estimate, not a hopeful one. On an unproven SKU the downside of a brief stockout is small and recoverable, while the downside of overbuying a product that does not sell is months of storage fees and trapped cash.
T. Brian Jones is co-founder and CTO of Inventory Hero. He leads the engineering behind its Amazon data pipeline, demand forecasting, and the AI platform that lets sellers talk to their live inventory, sales, and supplier data in plain language.
Plan the first reorder before launch and watch sales daily, because your lead time forces you to commit again while you still have only a few weeks of data. Use the earliest real sales velocity, adjusted for launch-period inflation from ads and promotions, rather than waiting for a clean signal you will not get in time.
Read article
To forecast demand for a new Amazon product with no sales history, build the estimate from analogs (a comparable product, category benchmarks, and your launch plan), bias the first order conservative, and plan to replace the guess with real data at the first reorder. The short version: the launch forecast is your best guess, but the first reorder, made on actual early sales, is the decision that really sets your inventory. Below is how to build the analog forecast, how much to order, and why the first reorder is the one that matters.
You forecast it from the closest signals you have:
An analog product. The best source is a similar SKU in your own catalog: same category, similar price, similar audience. Its launch curve and steady-state velocity are your starting template.
Category benchmarks. Competitor sales rank and review accumulation rate tell you the rough size of the demand pool. For a free read, eyeball the Best Sellers Rank of the top organic and sponsored listings for your main keyword: a stronger BSR across page one means a bigger pool. To turn BSR into a unit estimate, a third-party tool (Jungle Scout, Helium 10, and similar all publish sales estimates from rank) gets you a ballpark of whether you are entering a pool of dozens of units a day or hundreds.
Pre-launch signal. A waitlist, an email list, or early ad-click data gives you a small but real demand signal before day one.
If you are entering a brand-new category with nothing comparable in your own catalog, you have no analog, so lean harder on the category benchmarks, keep the first order deliberately small, and treat the launch itself as the test rather than trying to forecast your way to certainty. None of these inputs is precise. Combine them into a conservative first-order estimate rather than a hopeful one (any of the forecasting methods takes over once you have a few weeks of real sales), because the asymmetry of the bet favors caution.
On an unproven SKU, the two ways to be wrong are not equal:
Order too little and you might stock out briefly. On a new product with little rank to lose, that is recoverable and cheap.
Order too much and a product that flops becomes months of storage fees and cash trapped in inventory you cannot move, a far more expensive mistake on a SKU that has not earned its keep.1
So the first order should cover the launch period plus your lead time at the conservative estimate, no more. You can always reorder; you cannot easily un-order. For the inventory-staging side of a launch, see inventory for a product launch.
Your closest analog launched at about 8 units a day and settled near 15. You expect a similar product but stay cautious and plan the launch at 8 a day. With a 75-day lead time and a 30-day launch window you want covered before the first reorder lands:
Input
Value
Conservative launch velocity
8 units/day
Launch window + lead time to cover
105 days
First order
~840 units
That 840 is enough to launch and bridge to the first reorder without betting on the optimistic 15-a-day case. If real demand comes in hot, you reorder early; if it comes in cold, you have not buried cash.
The launch forecast is a guess made with no data. The first reorder is made with real data, and because your lead time is long, you have to make it while that data is still thin, often only two or three weeks of sales. Plan it before you launch:
Watch daily, not monthly. Early sales move fast and you cannot wait for a clean monthly number. Pull units ordered by ASIN from your Business Reports and look at the in-stock days only.
Adjust for launch inflation. Launch sales are often pumped by ads, coupons, and a ranking honeymoon. A 30 to 50 percent haircut off launch-week numbers is a reasonable starting point when you ran heavy PPC or a coupon, so you do not reorder to a peak that will not hold.
Reorder on the earliest trustworthy signal. A few weeks of real sales velocity beats the launch guess, even though it is noisy. The reorder point calculator turns that early velocity into a trigger once you have it.
Forecasting a new product with no history means building a conservative estimate from analogs and benchmarks, ordering enough to launch and bridge one lead time, and then putting your real effort into the first reorder, where actual sales finally replace the guess. The launch forecast gets you on the shelf; the first reorder sets your position. For the wider system, see Amazon inventory forecasting.