Skip to main content
Best practices for forecasting
S
Written by Sara Jaffer
Updated over a week ago

Configuring the forecast

When you are configuring your forecast, there are two main aspects you must consider:

  • Your lookback period

  • Any adjustment to the forecast

Lookback period

The lookback period governs how much sales and inventory history to take into account when producing the forecast. While on the face of it, it may seem obvious to use as long of a history as possible, in actuality you will want to choose a lookback period that considers your variant’s sales behavior.

If your product has a long history and sells evenly and consistently throughout the year, using a long lookback period is likely to be a good idea. Setting a long period means your forecast will be more stable.

If, however, your product is seasonal, considering the average sales across the previous year may not be a good indicator of upcoming demand. Instead, you may want to consider a shorter lookback period and use the adjustment field to account for seasonality.

Similarly, demand for trendy or fast-moving products may be more accurately predicted using a shorter lookback period.

Adjustment to the forecast

The forecast adjustment allows you to estimate how much more or less you will sell in a month compared to the calculated sales velocity. Adjustments are entered as a percentage increase or decrease.

The adjustment should be based on increases or decreases that haven’t already been factored in to the velocity. Therefore, the adjustment should also take the lookback period into account.

For example, if the lookback period is set to 14 days, then the forecast will change rapidly as the product’s sales change. Therefore you may not want to add an adjustment for the upcoming months.

However, if the lookback period is longer, you may want to use the adjustment to account for events such as seasonal increases or offers.

New products

Regardless of their lookback period, new products will have a short sales history as Inventory Planner only considers sellable days.

If a variant sells fast to begin with the forecast may be over-optimistic, though it may also be representative of ongoing popularity. Similarly, if a variant's sales are initially slow, the forecast may be lower than expected before it stabilizes.

In both scenarios, the forecast will stabilize on its own as the number of sellable days increases and Inventory Planner learns more about the variant's sales history. Therefore we advise users monitor new variants closely for the first 30 days, based on what they expect to happen and considering the influence of any marketing for launches, etc.

If the initial sales at launch are skewing the forecast too much, you can select a lookback period to exclude them once the forecast has stabilized.

Setting the forecast to zero

If you know you will not be selling a variant after a certain point, you can use the adjustment field to set the adjusted forecast to zero.

For example, say you are retiring Variant A at the beginning of August.

You do not want to include this variant in forecasts from August onward, as you are no longer going to be ordering it. Therefore to set the forecast to zero, you can set the adjustment percentage to -100%:

You should only have to make the adjustment for a few months, as once it stops selling you can adjust the lookback period to only consider days after your sales cutoff date, which will automatically update the velocity to 0 sales/day.

Note that using an adjustment of -100% to set the forecast to zero will always result in a zero forecast, even if the velocity changes.

Did this answer your question?