How Forecasting Works

Profit Hawk's forecasting engine drives every replenishment recommendation, urgency badge, and stockout date you see in the app. This page explains, in plain terms, how it works so you can trust the numbers and tune the inputs that matter.

The short version

For every product in every marketplace pool, Profit Hawk does this:

1. Calculates daily sales velocity from your historical Amazon orders, weighted by recency
2. Adjusts for seasonality if you have set a primary keyword on the product
3. Forecasts daily demand out 90 days into the future
4. Calculates safety stock to buffer against demand and lead-time variability
5. Recommends reorders and transfers by working backward from projected stockout dates and your days-of-stock targets

Each of those steps uses settings on the product. Get the settings right and the forecasts get sharp.

Sales velocity

Velocity is your average daily sales rate, measured in units per day. Profit Hawk pulls 18 months of order history from Amazon when you connect your account, then keeps it current with daily syncs.

Velocity is calculated for each timeframe: 7 days, 14 days, 30 days, 60 days, 180 days, and 365 days. The final velocity is a weighted average of those timeframes.

The weights depend on the velocity profile you set on the product:

The Sales Velocity table on the product detail page shows the calculated velocity for each timeframe along with the weight applied to each.

Seasonality adjustment

If a product has a primary keyword set, Profit Hawk pulls historical search volume for that keyword and uses it to adjust the daily forecast. Weeks with high search interest scale demand up; quiet weeks scale it down.

Choose a broad keyword that reflects the buying pattern for the category, not a specific listing term. "Yoga mat" captures category seasonality better than "extra thick blue yoga mat" because there is far more search history to draw on.

If you do not set a keyword, the forecast is a flat projection of weighted velocity with no seasonal adjustment.

Forecasted daily demand

The result of velocity + seasonality is a per-day demand forecast for each product, for each marketplace pool, projected 90 days into the future. You can see this on the product detail page as a forecast chart.

This forecast is the input to every other calculation in the system:

• The Days of Supply column on the Inventory Overview is calculated by walking forward day by day until current inventory plus inbound runs out at the forecasted demand
• The Stockout Date is the day that walk hits zero
• The Order Quantity recommendations are sized to bring projected inventory back up to your max days of stock target
• The Aged Inventory Analyzer and Storage Breakdown fee projections use this forecast for each SKU's sell-through, extended past the 90-day horizon by the keyword's seasonality pattern, so lifetime fees and removal recommendations reflect when a seasonal product actually sells

Lead times

Lead time is how long it takes to get inventory from your supplier to Amazon. It is broken into stages on the product:

The total lead time is the sum of the applicable stages. This drives the Order By date in supplier reorder recommendations: Profit Hawk works backward from the projected stockout date by the total lead time so the order arrives in time.

For warehouse-to-FBA transfers, Profit Hawk uses transfer-specific lead times (transit + prep) which can be set per warehouse-marketplace pair. See Product Settings for where to configure them.

Safety stock

Safety stock is the buffer Profit Hawk builds in to handle demand spikes and supplier delays. It is not a static number; it scales with each product's actual demand variability and lead time variability.

See Safety Stock for the full details on how it is calculated and how to tune the ABC classification that controls aggressiveness.

Reorder quantity

When the math says a product needs a reorder, Profit Hawk sizes the order using your min and max days of stock:

• Min days of stock (default 30) is the floor. When projected inventory dips below this number of days of supply, a reorder is triggered.
• Max days of stock (default 90) is the ceiling. The recommendation aims to bring inventory back up to this level after the order lands.

The raw recommendation is then rounded up to the nearest case quantity and to the MOQ if those are set, so you only see realistic numbers your supplier will accept.

When forecasts refresh

Forecasts recompute automatically:

• Daily, on a schedule, after data syncs from Amazon
• Whenever you change a setting that affects the forecast (lead time, days of stock, MOQ, velocity profile, primary keyword)
• On demand when you click Refresh Forecasts on the Inventory Overview toolbar

Recomputes typically take 10 to 15 minutes. The page will update automatically once they finish.

What can throw forecasts off

The model is only as good as its inputs. Common things that distort forecasts:

• Product was out of stock for an extended period. Sales velocity reflects what shipped, so a long stockout pulls velocity down. Profit Hawk does some interpolation to compensate, but extreme stockouts can still bias the result.
• Promotions or external traffic spikes. A one-week sale can pull the Responsive profile into thinking demand has doubled. Switch to Balanced or Stable for products with sudden, temporary spikes.
• Lead times that are wrong. A lead time too low pushes the Order By date too late. A lead time too high makes recommendations come too early. Update lead times after every supplier change.
• Missing seasonality keyword. If a product has clear seasonal patterns and no keyword is set, the flat forecast will be wrong on both shoulders of the season. The aged-inventory and storage fee projections also fall back to a flat sales pace for that product.

If a forecast looks off, see Forecast Looks Wrong for a structured way to diagnose it.

Related