Uncertainty is the biggest problem in
planning business life.
“Man plans, God laughs”, so says the Yiddish proverb. We are more connected, integrated and collaborated than ever before, yet demand remains unpredictable. And forecasts laughable.
“This is the new normal” – a phrase often used to capture this feeling. Operations planning and business decision-making – forevermore riddled with uncertainty.
In this context, where is the sense in planning? Best practices in lean manufacturing say that you connect production directly with the customer and get out of the way. This has lead to a supposed dilemma between execution-led lean and planning-based scheduling.
In most companies, sales and operations planning sits uncomfortably alongside lean. The tendency is for S&OP to gather momentum as it rolls downhill. Downwards through management hierarchy and time granularity until it collides with the actual demand.
The sales plan becomes a product plan. Product plan evolves to a production plan. By the time it hits production, it is a hard requirement.
Resolve this dilemma by thinking about two very different purposes for planning and scheduling: Prediction and Execution.
Plans are nothing; planning is everything.
Dwight D. Eisenhower
The function of planning is working out how to achieve certain goals. Scheduling determines when and in what sequence that activity will happen.
At a certain range, planning is purely predictive. Beyond the longest lead-time for material and resources, there is no execution required.
A plan is just part of a scenario. There can be many different scenarios, and the process of planning is valuable in exploring how those scenarios would play out.
The same goes for scheduling. Whereas planning asks what is required in capacity and materials to meet demand, scheduling shows what will happen when the plan is put into action.
Imagine that you are sitting in front of a control panel. There are a number of dials with settings that fix inventory policies, batch sizes, lead-times, minimum order quantities, production assets and headcount.
The combination of these settings will determine your business performance: too much min-max and you run out of cash; too long lead-times and you lose customers; batch sizes too small and you’ve no capacity left, too big and your inventory and lead-times blow out.
You have four choices:
Guess. Tongue poking out of the corner of your mouth. A wet finger in the air. Not very elegant, and quickly destined for trouble.
Set and Forget. Someone once worked out what the right combination was. Better leave them as they are and concentrate on firefighting.
Trial and Error. Change the dials one by one, check the result. It helps if you have a strategy and good visibility on what would happen.
Optimize. If you have all the data and a highly capable process, you could calculate the result.
In the real world we do not have all the data. And process variation is a fact of life. The best way to manage this complexity is through trial and error. This doesn’t mean you have to learn from costly business mistakes.
If planning is running with some kind of predictive model, then you can see the results that would happen. Before things actually happen.
Predictive planning takes actual or scenario sales data, runs it against material, process requirements and a set of business rules that determine how the operation will run. The resulting service levels, stock and cash-flow get reported from that scenario.
There is plenty of value in a predictive planning approach. Yet most planning systems seem to be driven towards execution.
ERP systems are transaction engines, looking for certainty and commitment. This is why it is often difficult to get visibility from an MRP system, without (or despite) expensive business intelligence and reporting tools.
A predictive planning model runs “what-if” scenarios. To do this, it needs to be both integrated with master data, but unhooked from generating transactions.
Your company may execute transactions based on a pull from Kanban, or from a push from the sales and operations planning cycle. If you are large or diversified, you probably do both.
Whatever the philosophy on push vs. pull, there is an increasing need to run a predictive model to ask those critical “what-if?” questions.
How To Build a Predictive Planning Model
All models are wrong, but some are useful.
Make things as simple as possible, but not simpler.
A predictive planning model does not have to be complex. You can see a simple example with this Capacity Planning Tool. Simple inputs are sales orders, bills of material, process routing, shift calendars and work centers. Outputs are a comparison of required load versus available capacity. Parameters could be start-dates, batch sizes, set-up times and overtime.
Each time this tool is run, you can see the result that arises from the inputs and the parameters. Change the parameters and run it again to see the effect. None of this means a commitment to a transaction that executes the change.
We use Excel to build predictive models because it is flexible and transparent. When we start with prediction in mind, it becomes possible to provide a much greater visibility on the future.
The same model can also be used to generate a signal for transactions. At close range, you will need to make a commitment to make a purchase or release work to production. So the reports then generate an action.
Planning systems have two purposes: prediction and execution. If your transaction system is able to give you good visibility on what-ifs and allows you to run scenarios then use it for both.
If not, then maybe you could build a predictive planning tool to complement the business system.
The Fast Excel Development method is geared up to build out these models quickly and easily. You can learn the foundation skills with our Fast Excel e-Learning tutorials, and implement it with the free Fast Excel Development Template.
Alternatively, we can help you get started quickly for a reasonable cost. Email us for an initial complimentary consultation.