LinkedInEmailWhatsApp

Sales forecasts reliability, how to improve it?

Discover ten best practices to improve the sales forecasts reliability within your company.

Why is it important to improve your sales forecasts reliability? 

The use of a Demand Planning Tool allows to forecast the future demand while analyzing sales historic data. It is important to point at that an under or over evaluation of this demand will have negative effect at several levels:

When the demand is under-estimated, the company risks not having enough stock to satisfy all its customers. This non capability of answering customer demand might lead customer toward competition, which can be translated in a market loss for a competitor. The shortage has also a cost that can be higher than just a lost margin due to the absence of sale. It will create a lack of confidence and downgrade the company images for its customers.  

In the opposite case, when the demand is overestimated, the company will have more stock to cover a need that might never come. To have too much stock means to immobilize resources. An overstock has a cost. A human cost, with the labor needed to move and tend to this stock. And a storage cost, with products that take spaces in warehouses. Finally, if the product has a limited life, the overstock can also become a waste, a net loss, as these products might be thrown away when the expiration date is reached.

A solution to reduce stock and improve service rate is to use an Advanced Planning & Scheduling tool such as Colibri in order to improve the current forecasts quality which will allow to better manage the stocks at the different Supply Chain levels.

How to improve your sales forecasts quality?

1 – Do the sales forecasts close to the final customer

Each element of the Supply Chain introduces a stock resulting from the downstream demand and the upstream replenishment. The objective is to avoid the Bullwhip effect, a common incident in the logistics chains. This effect consists in a tremendous amplification of demand variations while incrementally getting away from the final customer. Several factors explain this phenomenon:

  • A non-accurate and non-shared information between the different actors, with a lack of full disclosure.
  • Long lead time
  • And above all, a separation between consumption (actual demand) and production (actual activity)

2 – Work by exceptions: 

This concept means that it is better to work and spend time on what’s matter than dealing with specific cases, so small in term of volumes, that their impact would not be visible at a macro level.

Simples tools and methods exists to work by exceptions:

  • The ABC analysis, well known, allows to class products regarding their weight. This method is backed by the normal law saying that 20% of the products create 80% of the value. You can then apply management rules depending on the class. The objective is to focus in priority on class A products, the most important ones, where the demand has a strong impact.
  • Lifecycle management. A product will go through several steps in its life. In the mass consumption product ecosystem, pattern is usually the same when looking at historics: an introduction phase followed by a growth period at the beginning of the product life (launch phase) than a maturity phase and a declining phase. To have a criteria to describe the status of a product is a precious information allowing demand planning tools such as Colibri to implement specific management rules for each step of the workflow.

Having this information listed as attributes in Colibri allows to build work perimeters, alerts for each step of the workflow (standard in the tool), thus helping the person in charge of building the forecast plan.

 

3 – The implementation of a collaborative process

Users can bring qualitative information that might significantly improve sales forecasts quality. The demand planner is above all a hunter where the information is the prey.

To have a tool such as Colibri, natively managing collaboration is a genuine force to enrich the forecasts and thus take advantage of all the qualitative information available. It can be for the historics correction, to explain exceptional events, or for the enrichment of the statistical forecasts with other forecasts types (promo/event). The use of Excel is not ideal for this kind of process, as repetitive exchanges of files become too heavy to manage for the teams.

4 – Measure, Evaluate and follow the reliability: 

To improve the reliability goes first by the analysis of this reliability. Colibri archive up to 12 months of forecasts allowing to follow the forecasts quality at any level over a full year. To share regularly this indicator with the actors of the process will ensure a shared vision and allow to build a concrete plan to adjust important mistakes and thus reached the objectives set.

But improve the reliability also mean to not create bias with actions that might worsen the forecasts.

How to not degrade your sales forecasts?

A saying is that “a forecast is always wrong”. To have a good reliability means also not to add errors over the ones already there natively. To choose a robust model allows to limit additional errors.

1/ Prefer a robust model in time if a product is highly erratic in terms of sales.

A forecast hold, by definition, some error. It is better to tend to have a robust forecast over time, meaning a forecast that does not change completely one month to another. The choice of a statistical model is the first important step to ensure robustness of the raw forecasts which is the base for the enrichment phase.

2/ Work at an aggregated level ease the modeling and reduce the number of models to run (principle of the great number law) but by working at a too high-level risks are multiples:  

  • By working on a too aggregated level, you will blur the trend and seasonality effects, valued information for the modeling.
  • The split of forecasts at lower levels induce, by definition, an additional error. It is important to avoid having too much levels between the final and working level to not add errors.

3/ The forecast must drive the budget and not the opposite

To enrich the forecasts with a database about definite facts is a good practice. But, to set an objective as a value of the final forecast is not a good move to reach a high quality of forecasts, especially when you already know that those objectives are not reachable with the current available resources. The objectives either personals or corporates, should be compared with the current forecasts in order to analyze feasibility and implement an action plan to reach them if necessary.

4/ Do not work just with quantities

To valorize a forecast, allow to have a better idea of the impact modification can have. Colibri allows to manage sale price by perimeter, which allows to valorize forecasts and realize the effects linked to the price.

Sales forecasts reliability, to remember

In a nutshell, reliability is a key indicator to follow when working on your sales forecasts. The more the forecasts are reliable the less your company will risk being in an over stock or shortages situations.

Numerous actions can be implemented to improve this reliability but also to not deteriorate it with additional errors.  

In-fine, this indicator must be shared by all and match your company global strategy.

Watch our video about sales forecasts reliability

Watch the video