extract from "Business Excellence" by Phil Robinson
From chapter 2 - Sales and Operations Planning
4.3) Sales Planning
It is often as difficult to get Sales and Marketing interested in forecasting every month in a way that is useful to manufacturing as it is to get Manufacturing to believe the forecast when it is produced. Once the sales and operations planning process is operating, both sides soon find out that this is the mechanism that gets things done. The Sales and Marketing group will come to realise that the forecast is their way of "reserving" manufacturing capacity for their customers.
The only thing that you can be certain about when it comes to forecasting, is that the forecast will, in all probability, be wrong. The manufacturing challenge is to deliver the best possible service with the forecast information they can reasonably expect. The long term solution to the problem of sales forecasting is to increase flexibility and reduce lead times to the extent that you are able to react faster than your customers. The forecast then becomes much less important except as a long range or strategic planning tool.
There are few companies that do not find it difficult to forecast their future level of business. It is not uncommon to hear managers say they cannot use MRP because they cannot forecast. The reality is that a business will not run without forecasts. If there is no formal forecast, managers will make their own forecast. The purchasing people will buy what they think will be needed. Production people will maintain equipment they think they will need and even buy new equipment on the basis of their guess of what is needed. People will be recruited on the basis of what various people believe will be required. Budgets will be based on some kind of financial forecast and so on. Companies that say they cannot forecast will have a proliferation of local forecasts. The problem is these local forecasts are all likely to be different.
If we are planning a walk and the weather forecast says it will rain, we plan to take waterproofs. It may not rain but we will take the waterproofs. In a business sense, even though we cannot rely on the forecast it does not stop us from making plans based on these forecasts and executing them. As the forecast changes so we will adjust our plans. The advantage is that everyone is working to the same plan. We will have all our resources, people, equipment, material, capital etc. co-ordinated to meet the best estimate of what we think will happen. As our estimates of what will happen change, we adjust our plans in a synchronised way to meet the new circumstances.
If a business cannot avoid forecasting it make sense to have one agreed forecast that everyone works to. The most sensible place to co-ordinate the forecasting process is the sales and/or marketing department. If sales and marketing is separated, companies who have short term instability will generally find that sales can produce the most accurate forecast with an input from marketing for the longer term. Where there are long term issues, marketing will generally be able to produce the most accurate forecast but will need an input from sales for the short term. The breakthrough comes with the realisation that the fundamental reason for forecasting is to improve customer service by lining up all the resources to meet the best estimate of what our customers will want, when they want it.
Before a company can start to forecast it need to think carefully exactly what it is forecasting. The Wrigley company ships its chewing gum to distribution warehouses before shipping to shops. It found that different divisions were forecasting different things. Some divisions were forecasting shipments from the factory to the warehouses. Some were forecasting shipments from the warehouses to the shops. Some were even forecasting sales of gum to the public. For sales and operations planning, the forecast will be shipments from the factory to the customer or any distribution warehouses which are under separate control. If the factory is able to use the warehouses to buffer shipments, it will generally be shipments from the warehouse that will be forecast.
The other important consideration when considering what is being forecast, is to remember that the forecast should be the demand not orders accepted or shipped, even if the factory was unable to meet customer demand. The forecast accuracy has to be judged against demand and demand information, not orders accepted or shipped, has to be used as the starting point for future forecasting.
There are many other distortions in the supply chain that result in companies which are further down the chain seeing much larger swings in demand that are really taking place. For instance, whenever there is a "three for the price of two" promotion there is an obvious distortion in the flow of supply. Some distortions are less obvious.
Where there is strong pressure on monthly output figures there tends to be a month end ‘crunch’, ship anything and everything! This effect tends to ripple back through the supply chain. Credit terms are often such that a company will get the maximum credit by taking delivery at the end of the month. Quantity discounts based on the quantity supplied in one delivery will cause companies to save up orders to get their maximum discount and so distort the flow of product. Planning systems that increase safety stock based on recent history will amplify the swings. Long lead times and variable lead times will also add to the problem.
Most of the above distortions are avoidable. It is not difficult to set up performance measures, credit terms and a quantity discount structure that flattens the demand pattern. There are enormous benefits to be gained from a steady level of production.
For all these reasons, historical sales, more often than not, contain distortions. As many fluctuations as possible need to be ironed out of the historical figures before they can be used as a basis for forecasting. There are many forecasting techniques and algorithms that can be used to aid the forecasting process.
It is often possible to find a ‘lead indicator’ This is some related statistic which is available locally or nationally that can be used to predict future trends of your product. For instance, in the building trade there are good statistics on house starts but even mortgage rates may be a lead indicator.
If there are no lead indicators, it is often sufficient to look at the recent trend and project it forward. There are a huge number of sophisticated forecasting programs that can be ‘bolted on’ to an MRP II system that will do this.
Many companies have found that the best forecast is to take last year’s sales and adjust for any known variations. This is particularly applicable when lead times have been reduced to the extent that forecasting is not a critical issue.
The simple technique of Focus Forecasting has also claimed some impressive results. The technique was developed by Bernard T Smith and is described in his book called simply "Focus Forecasting". A simplified explanation of the focus forecasting technique is that you take historical sales figures that have had any distortions removed, and use the data to predict the last month or two of actual sales using as many forecasting techniques as possible (e.g. average, weighted averages, moving annual total etc.). The technique that gave the closest result is then used to forecast the next year. This process is repeated for each product. The whole process can be repeated every month.
The beauty of focus forecasting is that it is delightfully simple. As Smith says " A Complex answer is just used to hide the fact that the answer has not yet been found - complex answers don’t work". Whilst a sweeping statement, focus forecasting has certainly often produced better results than more complex, and often almost incomprehensible, mathematical algorithms.
Some companies will want to hold a formal meeting to agree the monthly forecast. At this stage the meeting will be confined to the Sales and Marketing personnel.