Best Practices for Connected Demand Planning

Posted by Alejandro Gomez on 03-Aug-2020 11:27:27


Headline Image 1-2

"Prediction is very difficult, especially if it's about the future" - Niels Bohr and Others


(6-8 MIN READ)

Since predicting the future is not that easy, let’s get started with a statement about the present: We live in the age of information.

Everybody – including planners and business leaders for the case in point – have access to a great volume of information and content on how to better plan the demand and increase the accuracy of the demand forecast. And I am sure that you, reading this, already have read a lot about it.

So, why another blog yet mounting on the existing pages that already cover demand planning? Because “prediction is very difficult, especially if it's about the future”. And predicting the demand is not only difficult, but pivotal for now-a-days demand driven organisations. The demand forecast ripples through the organisation and has major impact in Inventory Planning, Revenue Estimates, Budgets… and this relevance is what pushes heads of supply chain and other decision makers to find new ways to improve their predictions.

And in that attempt to find better ways to predict the demand in an ever-changing, fast-paced, complex world, is where planners dive into the abundant information available. But abundant information also brings along infoxication, which makes hard to grasp a good understanding of the matter and turn the information into actionable knowledge.

Different companies operate in different industries, commit different resources for planning and are in different stages in terms of planning maturity. So, what may work for some, for others perhaps it only contributes to increase the technology stack, lack visibility, and make business lose ownership of the process.

So, we are writing this blog to leverage the experience of Olivehorse’s practitioners (our Case Studies can be found HERE) and provide you with the best practices that ensure that your demand plan delivers an accurate, connected and meaningful Demand Forecast in an efficient way in the present and to be sure that it has strong foundations to be the bedrock for future enhancements.


"It is better to be roughly right than precisely wrong" - John Maynard Keynes


Let’s get cracking with a clean slate: predictions will always be wrong to some extent. No matter how sophisticated the demand forecasting method: It is impossible to produce accurate forecasts systematically.

That being said, the focus will be not only on improving the outcome of the statistical forecast and the ease of collaboration, but also on deploying streamlined, responsive and centralised framework owned by the business team itself.

This means designing a repeatable and auditable process that occurs in a single platform and that can be triggered at any time by demand planners (without depending on IT or third parties). This grants the capacity of planning and re-forecasting as events unfold in the real world, keeping plans always close to the execution layer and up to date, avoiding siloed planning and obtaining a single version of the truth, a single demand forecast shared across the board and impacting other business plans.


"A good forecaster is not smarter than everyone else, he merely has his ignorance better organised"  


So here is the deal, we have organized our experience and ignorance in a format aimed to provide you with a tested process that will bring great results and can be used as bedrock for future enhancements and improvements in your Demand Planning journey:

  1. Start by laying out a process: It is critical. Does not matter how you do it. Every organization must develop a demand planning process in order to be able to obtain a consensus planning that is time-efficient and reliable. Create a process like the one below. Assign clear tasks, user interfaces, expected outcomes, deadlines and responsibilities per step.

Image 1


Some key characteristics of a good process are:

a. Repeatable and time-proof: so it can be run at any, either running the complete process, resulting in a whole new Demand Plan (quarterly, monthly) or running it partially, adjusting the current plan with the latest signals (weekly).

b. Streamlined: so every contributor to the plan has clear tasks and a clear interface to work on when is time to run a new iteration. This allows to complete the job easily, so time can be spent analysing what matters rather than adjusting calculations in spreadsheets or collecting data from different emails and sources.

c. Auditable: What cannot be measured, cannot be improved. And that applies to Demand Forecasts, too. Every time a new forecast is produced, the outcome has to be recorded so we can compare the different plans against the actuals, measuring the accuracy achieved by different contributors per product and location, achieving continuous improvement.

d. Ubiquitous / Cloud: being cloud means being ubiquitous, so people all through your supply chain and across the board can input their insights and demands signals in real time and from any device.

e. User-friendly: so contributors not only can do their job anytime, everywhere, but they will also find it easy to do, so they will be eager to adopt the system to input their data.


2. Foster collaboration: a good demand plan must include the view of all these teams involved in demand generation and the ones serving the customer. And it must be aligned with other forecasts, such the Financial plan, after all, your demand probably accounts for most of your P&L’s top line.

But “Collaborative Plan” does not mean “one-size-fits-all Plan”. It Means creating proper user interfaces that allow each team to speak their own language and yet contribute to bring more insights to the plan, rather than just sticking a number in it.

For instance, under these lines there is an example of a dashboard designed for Marketing to enter their planned promotions, capturing the duration, the budget, the expected ROI and expected demand uplift, among other key parameters.

Image 2

Besides the marketing input, other stakeholders in the Demand plan are:

a. Sales intelligence: Account Managers can bring into the plan the sales intelligence, weighting in the latent and emerging demand.

b. Finance alignment: include financial forecasts into the consensus forecasts to link the financial plan to the operating plan.

c. Customer collaboration: seamless integration of the insights obtained at store level with the plan, resulting in an ongoing short-term demand sensing that significantly improves the accuracy of the near-future demand forecast. Each industry has its own signals for demand sensing and these signals can be qualitative and quantitative.

The end goal is to provide all the relevant data to the ones in charge of the forecast, in a transparent way, as per shown on the picture below. By providing last year’s actuals along with the statistical baseline and the vision of the different departments (and their past accuracy), planners are in the right position to produce the best possible consensus planning.

Image 3

3. Data & Statistics: as the volume of data available to be analysed grows, companies not only need to be able to store but to analyse it in a timely fashion in order to obtain the best possible insights, for instance:
Image 4
a. Finding the right level for forecasting: after loading actual historic data, some SKUs/location combinations might have a rather variable or intermittent demand throughout the year. By determining the coefficient of variance of the demand of the product/locations, we will be able to decide whether will be more accurate to forecast certain products at Brand or Product Group, in order to reduce the noise and increase the forecast accuracy.

b. Find the most convenient forecasting algorithm: by setting up calculations for several different time-series forecast [LINK], after loading the historic data, we will be able to find the forecast that fits the best the historical demand for every product by measuring different errors, as per shown on the picture above.

c. Forecast at SKU / Customer level: This level of detail does not only enable planners to slice and dice the forecasted demand for better insights, it also fosters the collaboration by producing a prediction measured in the same units of measure as the ones used by Marketing, Account Managers or Points of Sales in their plans. Planning at that granularity will allow to use the demand forecast plan the packaging needs.
d. Slice and dice the data: for instance, slice the demand per sales channels. Eg: drinks, brewing, food industry often categorize the demand in channels such “Retail” (Tesco,…), “On Trade” (Wetherspoons, Marston’s…) and other small groups of restaurants and stores. Each channel has its own particularities and we might want to run independent forecast for each channel.
Image 5

We really hope that the guidelines above help to guide or reassure you in your demand planning journey. and

…Book your free taster session now!

We would love to meet you and have a chat on what can Anaplan do to optimise your inventory or boost your Supply Chain Planning.

So, feel free to Book your FREE Taster Session now where we will spend half a day with you, learning all about your business and demonstrating the value that Olivehorse’s solutions in Anaplan can bring to your company. 

OH Anaplan Multi Echelon Safety Stock Blog March 2020 Meeting

The consultants from Olivehorse will meet you to discuss how Anaplan can improve your Demand, Inventory Planning, S&OP or any other planning or execution process.

A Gomez

Alejandro Gomez is Anaplan Practice Lead and Master Anaplanner at Olivehorse. He has been working with Anaplan for 6 years now and participated in the design and implementation of a several projects across industries and business lines.

Together with the Anaplan team, Alejandro makes sure to capture all the calculations and requirements gathered by the Inventory Team in an Anaplan model that complies with all the Best Practices.

Read more on: Inventory Optimisation, Anaplan