Choosing an optimal forecasting system for your retail company

Home » Blogs » Choosing an optimal forecasting system for your retail company

Forecasting is critical to any retail company. Without having a forecasting process in place, it is near impossible to have the right stock on hand.

Too much merchandise in the warehouse means your inventory is collecting dust, and not enough on the shelves can hurt your brand image. Such a fine line between success and failure!

In this blog, we take you through our process of finding the right forecasting system for your retail business.

For starters, what is forecasting?

Forecasting uses data and insights to accurately predict demand at a granular level. This analysis allows retailers to know how much stock to have on hand at a given time. However, if not done correctly, the results are just the same as putting your finger in the air to find the best guess. That’s where the power of forecasting systems comes into play.

How we choose an optimal forecasting system

An optimal forecast takes into account:

Sourcing and replenishment cadences

Geographical and socio-economic variances in demand

Out of pattern disturbances (ex: Out of stock conditions caused by COVID, severe weather events)

Business Goals and overall profitability

From these considerations, we can define the forecast timeline, forecasting accuracy, scorecards, and overall return on investment.

Determine a forecast timeline

A forecasting system is only successful if the retailer has a strategic plan for a forecasting timeline.

Rather than focusing on the exact promotion history, a successful forecasting timeline anticipates changes in demand that factor in essential business decisions: promotion strategy, vendor ordering, and seasonal items.

Calculate forecasting accuracy

It is important to use metrics when comparing forecasting accuracy against competitors.Note that forecast accuracy does not rely on one metric, rather a combination of several metrics depending on the business strategy.

Here are the recommended metrics to use based on different business strategies:

Forecast Bias: (Forecast – Actuals) / (Forecast + Actuals)	This metric can show if you are over-forecasting or under-forecasting.
Percent Absolute Error (PAE): Mean Absolute Percent Error (MAPE):	These two metrics provide a clear understanding of the error when comparing forecasts. Note, they can be sensitive to outlier forecasts on relatively unimportant partitions of the business. Example: An over-forecast of a product that sells one or two units per week can drive very high MAPE and PAE values which distort the overall accuracy of the forecasting system.
Normalized Weighted Root Mean Square Logarithmic Error:	This metric provides additional information on the quality of the forecast that is less sensitive to outlier error contributions. It also gives a more holistic view of the error distribution when comparing forecasting systems.

Calculate accuracy by comparing naive forecasts to advanced forecasts

It is advisable to compare an advanced forecast with a naive forecast (an estimating technique) to understand the true value.

Naive forecasts rely on broad averages and are only used to compare with sophisticated forecast systems. If the contending forecast systems can more accurately model the seasonal and promotional response in the system, they will have lower MAPE and Logarithmic Error.

Create forecasting system scorecards

Forecasts for one system might be very accurate in a subpart of the business but perform poorly in another. For example, the forecasting system might perform well for promotions with a high response, but overall forecast highly seasonal, unpromoted items.

To generate a comprehensive insight into which of the competing forecast systems is most accurate, we recommend creating a scorecard.

The key components of a scorecard:

Determine appropriate weighting factors aligned with the business strategy and the nature of the products being forecasted. This will avoid overweighting items such as slow sellers, out of season, or discontinued/clearance items that disproportionately impact the metric.
Leverage statistical formalism to determine if the difference between the forecasts is mathematically important.
Ensure that all categories that make up the metrics align with the overall business strategy

All of the contributions to the scorecards are tallied and evaluated along several essential dimensions of the business. The forecast system with the lowest overall point total is determined the most accurate among the competing systems.

Compare ROI

Discovering the return on investments (ROI) for each forecasting system is just as important as creating scorecards. Understanding the ROI allows us to understand the business impacts, and calculate the costs of moving to a new system.

To calculate the ROI on each forecasting system, we combine both business and industry knowledge. This includes inventory changes, sales quantity changes, sales revenue changes, and more. Monetization of these effects directly impacts inventory costs, weekly revenue and lost sales, and percentage average error (PAE).

Takeaways

Forecasting is a crucial business function for any retailer. However, without an accurate forecasting system aligned with the business strategy, the controllable risks to your business and bottom line can be substantial.

At Cognira, we believe that our methodology provides critical input and confidence into the decision process for retailers.

About Cognira

Cognira is the leading artificial intelligence solutions provider for retailers. Cognira is passionate about helping retailers unlock valuable, transformative business insights from their data.

We know retail. We love data.

To learn more, check out our website at cognira.com or contact us today to get started.

Subscribe To Our Newsletter

Get the latest updates about retail technology

Share This Post

+ More To Explore

Forecasting