This article covers calculations used for the Restock Report.
Note: You must be subscribed to Insights to have access to the Restock Report.
Forecasting
The Restock Report is designed to interpret and use the best possible forecast model for each MSKU tracked on it. It decides this based on the following criteria:
- The total number/volume of sales for the MSKU
- How long you have been selling the MSKU
- Whether there are any predictable trends in past sales overtime (e.g. sales for an MSKU are higher during certain months of the year).
If available, up to the last two years of sales history for an MSKU are used to compute sales forecasts and replenishment recommendations.
Our forecasting models use an 80% Confidence Interval. This interval was chosen because there is a trade-off between higher confidence and precision- a higher confidence would give too high of a range of data points.
We use error metrics against our calculations, and the forecast method with the lowest error metric for each MSKU is assigned automatically.
The Forecast Model will show as Not Enough Data if there has not been at least three days' worth of orders over the last two years for an MSKU.
Linear Model
Formula: Y = a + bX
Y is your existing (historical) sales for a given day (based on quantities ordered)
a is an average of your existing sales (based on quantities ordered)
b is the upward or downward trend of sales
X is the existing sales date or the date to predict sales for
The Linear model is selected when an item's sales history provides data that shows a trend with no fluctuations. This forecast requires the least amount of historical data for an MSKU, so you may see this model used to forecast items that have not had many orders in a wide timeframe.
With this model, you will see a steady, straight line when viewing the Sales Forecast graph for an MSKU.
Example from the Sales Forecast column:
Example from the Sales History & Forecast graph:
Moving Average Model
Formula: MA = (Y[1] + Y[2] + Y[3] + . . . Y[N]) ÷ W
Y is your existing (historical) sales (based on quantities ordered)
N is a given day
W is the window of time being used for the calculation - we are using a window of 5 days
The Moving Average forecast is selected when an item's sales history shows a trend with fluctuations. It requires more data than the Linear Model and compares the mean of your most recent orders against historical orders averaged together to better account for these fluctuations in sales.
With this model, you will see a fluctuating line when viewing the Sales Forecast graph for an MSKU.
Example from the Sales Forecast column:
Example from the Sales History & Forecast graph:
Seasonality Moving Average
Formula: This model uses the above formula for the Moving Average model, then applies steps of logic to reach a result that takes the seasonality of sales into account:
- Find the indexes for seasonality, which are calculated for each month of a year, so there are 12 total indexes in a given year.
- The moving average is calculated for each of these 12 months based on the combined sales per month.
- Next, we find the centered moving average (mean of moving averages in the window, which is 3).
- We then calculate the standard deviation (on average, this is how far each data point is from the moving averages gathered above) from the centered moving average.
- Then we calculate the standard deviation for each month and average them (i.e. all for every March of every year, or if there is only one data point in March, then no averaging is needed) to get the index for that month.
- Find the line of best fit (using our linear calculation) for existing sales (quantity ordered) and combines the predicted value with the seasonality index for the month of the given date.
The Seasonality Moving Average forecast is selected when you have at least 13 months of sales data that follows a predictable trend. Comparing this year's predictions with last year's sales data reflects seasonality and irregularity dependent on the calendar months of the year.
An example of an item that may have this model assigned is a meltable ASIN, which can only be shipped FBA during certain months of the year.
With this model, you will see a line that fluctuates during certain parts of the year when viewing the Sales Forecast graph for an MSKU.
Example from the Sales Forecast column:
Example from the Sales History & Forecast graph:
Outliers
If there is a sudden spike (high or low) in sales for an MSKU, we identify this as an Outlier. An example of this could be if an item suddenly trends on social media and becomes high in demand for a short period of time.
Since these instances can cause a forecast to consider sales figures that are not typical for the MSKU and potentially skew data, they can be Included or Excluded from your sales forecast.
To Include or Exclude Outliers, click the toggle towards the upper right of the report or graph:
When the Outlier toggle is turned Off to Exclude them, InventoryLab will replace Outliers in the sales data for our forecasting calculations with the average of the last 7 days prior to the Outliers being removed. This provides a more realistic trend that isn't too high or low because of Outliers spiking the average.
Safety Stock
Safety Stock is the amount of stock recommended to have on-hand for an MSKU in case you start selling more than usual. It impacts the Reorder Point for an item.
The Days Remaining w/ Safety Stock is how many days of Safety Stock are remaining before you need to replenish. The calculation for this is:
Days Until Reorder Date = (Current Inventory Qty – Reorder Point) / Avg Daily Sales
Reorder Point (this is in-stock quantity, not a date) = Lead Time Usage (Avg. Daily Sales * Lead Time) + Safety Stock. Average Daily Sales – the average daily sales based on the past 30 days
Reorder Point without Safety Stock = Avg. Daily Sales * Lead Time
Safety Stock = Avg Daily Sales x Lead Time
Average Daily Sales = Avg. Daily Sales of the past 30 days. If there are no sales in the past 30 days, this is the Avg Daily Sales of All Time
To Include or Exclude Safety Stock amounts in your Replenish By dates, click the toggle towards the upper right of the report or graph.
Est Replens Qty
The Estimated Replens Quantity is the estimated number of units recommended to replenish for an MSKU depending on the Est Replen Period being viewed.
Qty needed to cover X days of sales:
- Sum the amount of forecasted sales between today and Lead Time days.
For example, if Lead Time is 10 days, sum days 1-10 of projected sales. (This is rounded up.) - Sum the amount of sales forecasted for each day of expected coverage after days of Lead Time.
This is the Qty Needed to Cover.
For example, if the expected days of coverage is 30 and Lead Time is 10 days, sum days 11-40 of projected sales. (This is rounded up.) - Take Fulfillable on-hand (from your Inventory Detail page) and subtract the sum of projected sales per day for today through Lead Time.
This is the Remaining Qty at End of Lead Time.
For example, if Lead Time is 10, sum days of 1-10 projected sales. If the sum is 5 and Fulfillable on-hand is 20, subtract 20-5.
If the remaining qty at end of Lead Time is greater than 0, Estimated Replen Qty = Qty Needed to Cover - Remaining Qty at End of Lead Time.
If the remaining qty at end of Lead Time is less than or equal to 0, Estimated Repln Qty = Qty Needed to Cover.
If the remaining qty at end of Lead Time is greater than qty needed to cover, Estimated Replen Qty = 0.
The Bottom Range is calculated by taking the:
Qty Needed to Cover - 80% Forecast Model Confidence Interval + Safety Stock (if applicable)
The Upper Range is calculated by taking the:
Qty Needed to Cover + 80% Confidence Forecast Model Interval + Safety Stock (if applicable)
If there is no range, this means that the Upper and Bottom range formulas calculated the same number.
Est Replens Cost
The Estimated Replenishment Cost/Unit is the estimated cost of the units that are recommended to replenish for an MSKU.
The Bottom Range is calculated by taking the:
Most recently replenished Buy Cost X Bottom Range Est Replens Qty
The Upper Range is Calculated by taking the:
Most recently replenished Buy Cost X Upper Range Est Replens Qty
0 comments
Article is closed for comments.