Updated to reflect changes in Amazon fees, tariffs, and inventory planning.
By this point in Q1, many brands have already determined their sales goals for this year. Hopefully you’ve looked at historical data to understand how your brand performed last year, and have used that data to create a roadmap for where you want to go.
One key component of this is creating an accurate unit forecast for each of your SKUs. This can be a complex process, especially if your brand has a large catalog, but it’s absolutely critical to your success!
We’ve created a complete guide to unit forecasting on Amazon, so your brand can see continued success in 2026!
What is Unit Forecasting on Amazon?
Unit forecasting is a form of demand planning where brands can predict how much inventory they’ll need for a specific SKU. Obviously, this is critically important to avoid over- or under-stocking fees in FBA, but it’s also a way for you to accurately predict sales for the upcoming week, month, or year.
Why Forecasting on Amazon Often Fails
Forecasting might seem easy, but there are many reasons why your Amazon forecasts might be off. Here are the three most common reasons our team sees forecasting fail on Amazon.
Treating Last Year as Truth
The first reason your Amazon forecast might fail is because you’re treating last year’s data or sales as the one source of truth for next year’s forecast. Of course, this is a great starting point—you need to base your forecast on historical data—but there are many factors you also need to take into account. Stockouts, price changes, and high return rates need to be accounted for in your forecast.
Within Kapoq, you can manually remove a lot of returns or out-of-stock time from your forecast, so you can see what your sales for this year would look like without those extenuating circumstances.
Ignoring Larger Impact of Stockouts
You don’t just lose sales during the out-of-stock window — you lose rank and visibility, competitors take over your keywords, and it can take your weeks or even months to claw back. All of these variables will affect your forecast for the upcoming year. The downstream mechanics of being out of stock make forecasting and inventory planning unforgiving.
To combat this, you can build buffers and earlier replenishment for hero ASINs, and monitor OOS risk more frequently. These small adjustments can make a big difference in your forecasting.
OOS impact on Amazon = lost sales + lost rank + higher ad costs + slower recovery
Not Accounting for Sales Curve
The second mistake that people make in their unit forecasts is not accounting for sales curves such as seasonality and demand shape. This sales curve determines which SKU grows by how much, in which month of the year, allowing you to account for periods where your sales might naturally be higher.
If your products see increased sales during Cyber Weekend and Prime Day, you’ll need to account for these tentpole days as well. Including these in your unit forecast can also help you accurately predict how much inventory to send to FBA and AWD.
Poor Data Quality
Perhaps the biggest culprit of incorrect forecasting is poor quality data. If your process for forecasting is manual and time consuming, you’re likely skipping over key details and missing out on that granularity in your forecast. Especially for new products, establishing reliable sales curves is impossible without historical data.
Within Kapoq, we need at least 3 months of historical data in order to predict a sales curve, but 8–12 months is preferred to get the most accurate data, especially for new SKUs.
Not Accounting for External Factors
One of the major reasons forecasts are missed is because brands aren’t taking into account what happened in their business outside of Amazon. For example, maybe your brand expanded into a big-box retail store like Dick’s Sporting Goods, or maybe you changed your pricing on your owned website.
In order to ensure your forecasting is taking this into account, you should include non-Amazon events in your forecast assumptions. This could include retail launches, influencer hits, PR, price harmonization changes, DTC promo shifts, and the like.
Keep in mind that you always have to take MAP pricing into account; external pricing can cause issues if a retailer undercuts the price on Amazon.
How to Improve Your Forecasting Process
The good news is that despite all of the challenges you might face in the demand planning and forecasting process, there are a few easy ways you can improve your unit forecasting in order to accurately predict sales curves!
Adjust Constantly
Forecasting is not something you do once in January and lock in for the year. Forecasts must be adjusted constantly, be judged against actual performance, and take into account both predictable and unpredictable changes throughout the year.
For example, you might need to exclude a week-long period from the previous year when a freak storm caused people to buy more batteries, or a whole shipment of your inventory got lost in the ocean from China, causing you to stock out for months.
Most brands forecast monthly in order to get the best, most accurate data, but some brands with long manufacturing times will forecast weekly. You need to make the decision that’s the best for your brand, regardless of what industry best practices say.
Your unit forecasting review frequency should be faster than your replenishment lead time. If you update slower than your lead time, you’re flying blind. During your peak season, you need a tighter forecasting cadence.
Utilize Technology
It might seem obvious, but having a great technology partner can make the unit forecasting process a million times easier. Whether you’re doing a strategic or trend forecast, the right tech can shave hours off the prediction process with accurate automations.
Kapoq makes this process easy by allowing you to build your forecast in seconds. All you have to do is download the unit forecast workbook. We’ve already pre-determined the best sales curves for each product in your catalog—all you need to do is manually review, account for other variables, and add your growth factors based on your goals for the year.
You can then upload that forecast into Kapoq and measure Actual vs. Forecasted performance to get a sense of where each forecast falls and what feels best for your business.
Follow This Monthly Mini Playbook
A good base trend forecast is around 70% accuracy, but getting to 90-95% accuracy requires true bottoms-up work and review loops. The most important practice to reach a higher level of accuracy is to regularly review your forecast vs. actual and diagnose why your forecast was misaligned (channel expansion, promo differences, retail assortment, etc.).
Each month, you should conduct an actual vs. forecast variance review. During this review, you should tag variances by cause (OOS, promo, ASP, channel, competitor, reviews) and then feed those learnings into next month’s strategic adjustments. This will allow your forecasts to improve over time!
Trend Forecast vs. Strategic Forecast
As you begin to dive deeper into unit forecasting on Amazon, you’ll need to understand the difference between a trend forecast and a strategic forecast.
A trend forecast is projecting recent history forward — basing your forecast on what happened in the same period the year before. This is a great baseline forecast, and is what most brands do when they’re forecasting.
A strategic forecast, on the other hand, layers in elements that will change — ASP, promos, channel shifts, inventory events, etc. It’s basing your forecast on past decisions and known future changes, understanding that forecasts are not 1:1 from the previous year. Whether you had a new competitor enter your market, a change in reviews, or a new retail rollout, all of these impact your strategic forecast.
How Demand Transfer Impacts Forecasting
Demand transfer is the concept of matching your customer’s desire with what you currently have in stock — even if it wasn’t what they were originally looking for.
For example, if you think of a typical retail store, you might have a customer stop in to look for a specific model of bike in black. If you’re out of stock of that model in black, they might just buy the blue version you do have in stock. The demand for one child ASIN transfers to another very easily in store.
On Amazon, however, demand transfer often isn’t as smooth. On your product detail page, your customer can see your competitors’ listings and transfer their demand there, rather than to a different child ASIN of yours.
Additionally, Amazon may keep surfacing your out-of-stock version on the search results page, which leads to a bad customer experience where customers click to your PDP only to realize the option that was pushed to them is out of stock.
It can be difficult to estimate how demand transfer on Amazon will impact your unit forecasting, but there are a few key ways you can estimate its impact. By looking at historical sibling-ASIN behavior, variation share, price parity, reviews, and category norms, you can begin to understand how being out of stock of a certain variant affects sales of other variants.
1P vs. 3P Forecasting: What Changes
Forecasting what your customers will buy is not always the same as what Amazon will buy from you. This is the key difference in 1P vs. 3P forecasting strategies. Amazon’s inventory position and weeks of coverage change the purchase order behavior.
On 3P (or FBA), you control your replenishment decisions; but on 1P (or Vendor), Amazon’s ordering logic and coverage reigns supreme. Being able to predict what Amazon is likely to order weeks ahead of the PO date is a huge challenge for 1P sellers. You have to be prepared to ship if Amazon orders, but that order is not guaranteed.
To prevent stockouts with FBA, it’s recommended to keep 4-6 weeks of coverage to ensure you always have enough inventory in stock. This should increase to 6-8 weeks during tentpole days or peak seasons.
Book your free demo of Kapoq today to set yourself up for success in 2026.
