Flowcasting Implementation - Lessons Learned

As loyal readers know, we’ve recently completed the implementation of the Flowcasting process for the Canadian hard goods retailer, Princess Auto. As a reminder, at a high level, here’s how their planning process works using the Flowcasting process and solution:

  • A single, time-phased forecast of consumer demand, at item/store level drives all planning activity from store to supplier
  • Long term visibility is calculated using the concept of dependent demand, shared throughout the organization and automatically translated into the language of the business (units, cube, weight, dollars)
  • Everyone is working and planning to a single set of numbers and a true model of the business exists inside the Flowcasting planning system
  • Transfers and orders are planned in advance, but are only committed to at a single lead time, regardless of the volume (i.e., promotions orders are created at the same lead time as regular orders)
  • Each day, based on any change, the entire supply chain is re-synchronized so everyone is working with the latest information

Recently, I had the pleasure to give a talk with them at the Supply Chain Management Association Ontario annual conference, and it afforded me the opportunity to grab a couple of beers with Ken Larson and to talk about lessons learned.  Mr. Larson was the Business Sponsor of Flowcasting during the design and implementation, and played a key role in its success.   Given a number of retailers are likely to embark on a Flowcasting journey, here then, are our lessons learned.

It’s about Behavior Change
Flowcasting is first and foremost about people, process and changing behaviors – or as Ken likes to refer to it, “the mental model”.  While you cannot implement the process without a system, paradoxically the more time and effort you spend addressing people and process, the better.

The approach we used was an approach that we’ve honed over the years and has its seeds originating from the Proven Path approach that’s been successfully applied thousands of times by the Oliver Wight Group.  Over the years, we’ve improved on this approach and added details and activities to help ensure that people come to the conclusion themselves that the change is a better way of working.

It’s an approach that requires lots of time and effort on education, process labs, pilots and coaching.  I can still remember Ken initially questioning the amount of education we were planning.  As we were finishing up the rollout out he said to me, “you know, you can’t do too much education”.

Big Leaps aren’t that difficult
Flowcasting looks like a massive change and leap of faith for any retailer.  While it’s true, there is a lot to change for most retailers, our experience is that it’s not nearly as difficult as some might think.   That’s because the process and change are intuitive and natural.

After all, the Flowcasting process plans the way we do business.  We create the forecast only at the store level, because that’s where sales happen.  The DC’s demand and plans are to satisfy the store.  The supplier schedule is to satisfy what the DC’s need.   The lesson here is that even a big change on the surface doesn’t have to be that difficult, especially if it is natural and makes sense.

Process harmony is more important than collaboration
Read any literature regarding supply chain planning and it’s very likely collaboration will surface.  One of the most counter-intuitive lessons learned during the implementation was that, for a retailer utilizing the Flowcasting approach, collaboration was not nearly as important as process harmony and integration.

First, let’s address collaboration and its cousin, consensus.  Since the planning process is driven by a forecast of consumer demand, by item/store, the ability for different groups to improve the forecast by adding valuable knowledge is quite limited.  The retailer holds all the information that are the essential drivers of consumer demand and what we determined is that other groups, particularly the manufacturers, either had some of the same information, or were missing large pieces of information entirely.  It would be almost impossible for them to improve the consumer demand forecast.

Now, suppose we decided to go through the effort to share with them all the same information – POS history and all the attributes that drove promotional sales, for example.  Why would the supplier derive any better forecast using the same information?  The same thinking largely held true internally between the Demand Planner and the Buying Team.  There’s an old saying “in business when two people always agree one of them is unnecessary”.  We think a similar axiom holds true in forecasting consumer demand.

What we discovered, and believe it will be true for most retailers, is that a retail demand planner has the information and knowledge to determine the consumer demand forecast.  Sure, in some cases for things like limited history promotions, the demand planner may need to collaborate and come to consensus internally with the Buying/Merchandising/Category Teams, but this is proving to be an exception rather than the norm.

What’s more important is what we call process harmony and integration.  The demand and supply planning processes do not work in isolation.  They need to integrate closely with key Merchandising processes like Assortment Planning/Line Review and Promotions Planning, to name two.   These processes need to work in harmony and to a heartbeat that works for the business.

For example, new products cannot be introduced and planned violating lead times required to get the product on the shelf.  The same is true for promotions planning as this process needs to be performed far enough into the future so that all parties can plan to fulfill the demand.

Harmony and integration are achieved by co-designing the Flowcasting process with these processes and ensuring that the entire company is educated and understands these key process linkages and timelines.

Software is important and should be largely invisible
As mentioned above, it really is all about people and process.  That of course, is only true as long as you have software capable of supporting the Flowcasting process.   Using the wrong software will be the kiss of death to your efforts, as most of the time will be spent jury-rigging software not suited to the job.

As an example, I know of a number of retail companies that have attempted to enable the Flowcasting process starting, of course, with a forecast of consumer demand at item/store level.    However, since the solutions didn’t have the capability to forecast at the item/store level, attempts were made to forecast at a higher level and then spread this forecast down to each store.  Unfortunately this hasn’t worked well and it dramatically complicates the forecasting process.

The solution we used was designed for the job and as a result the focus of the implementation was a polar opposite to most implementations.  We spent most of our time on process, people, educating, and changing the mental models of people, rather than on the software and integrating it to their legacy ERP system.  And, isn’t that the way a Flowcasting implementation, or any behavior/change implementation, should be?

Our lesson is simple.  Find a solution that works for the task at hand and then the implementation can be largely focused where it should be. What we found is that the software was largely invisible to the effort – that is, not a lot of effort, focus, discussion or issues were as a result of the software.

I’d like to thank Ken Larson and the entire Team for their work, commitment and sponsorship to help make the Flowcasting implementation at Princess Auto a success.  We’re very hopeful that you take these lessons to heart as you embark on a Flowcasting journey, or any change for that matter.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>