Costume mustache and classes

SAP Demand Sensing: A Supply Algorithm in Disguise

Here’s a breakdown of where demand plannning fits in most organizations and how it can deliver incremental value to your supply chain by reducing costs and delighting customers.

It’s always bothered me that demand sensing is in the SAP IBP Demand Module.

It has “demand” right there in the name, so I guess I shouldn’t be all that surprised.  However, I think it’d be more fitting to include in the supply and response side of the IBP house, since the actual results lend themselves more towards responding to the market as it evolves (as compared to forecasting trends out in the planning horizon.) 

Demand sensing is a relatively new concept to most planning organizations. 

 

So where does it fit?

 

Who should manage it and design it? 

 

And how do you measure this thing anyway? 

 

Glad you asked!  Based on our experiences with demand sensing, here’s a brief breakdown of where it fits in most organizations and how it can deliver incremental value to your supply chain by reducing costs and delighting customers.

Reaction is not a Forecast

The window for demand sensing is in the 6-8 week range.  Most forecasting metrics that measure forecast accuracy of a demand planner are in a 1-3 month lag horizon.  Translation: Demand planners are measured on what they call as a forecast 1-3 months from the actual event.  When demand sensing takes over, the horse has left the barn for demand planners, the die has been cast and nothing can change their metrics that impact performance, and their related bonuses.

 

Demand sensing will pick up the baton and fine tune what the demand planning team has prepared.  With guidelines and boundaries around the amount of change tolerance allowed to the existing forecast, planners can review the exceptions and let the demand sensing application adjust as it sees fit.  The algorithm in IBP uses some machine learning components to review incoming sales orders as well as up to 8 other correlating factors to adjust the demand signals within that range at a daily level.

Daily splits benefit deployment- not production

Deployment involves getting the right product to the right place at the right time.  From my experience, this means a daily run of deployment solutions.  However, the effectiveness was not quite where it needed to be due to the inaccurate split of weekly to daily buckets

 

This is a hidden gem of demand sensing. 

 

It uses a form of machine learning to understand how to split forecast from weekly to daily much more accurately than an even split or a fixed ratio.  It senses patterns in the historical demand signals and uses that history to determine if Friday this week will be more than next Monday, and by how much.

 

Accuracy at this level of detail is a game changer in the supply chain space.  The capability to accurately predict demand at a daily level was a bridge too far for previous applications.  IBP running a machine learning type algorithm on a HANA platform is the difference maker.  Demand sensing is better due to the technological breakthroughs, and the IBP offering from SAP makes it accessible to anyone.

Don't predict. React.

Demand sensing is meant to fine-tune (not re-create) the forecast.  As a planner you set boundaries and parameters around how much the forecast can change on “autopilot”.  The efficiency of this planning process allows for planners to spend their valuable time on the exceptions or calculations outside of the guardrails and figure out if the change is a trend or an anomaly.  This real-time adjustment capability reacts to forward-looking real-time data.  Actual sales orders, forecast trends, point-of-sale data.  These data points show what is actually happening- NOT what happened during this week last year.

 

The correlations that the demand sensing algorithms pick up find opportunity, and make adjustments that would take an army to figure out manually.  And realistically they probably just couldn’t do it fast enough anyway.  This allows for a self-healing supply chain; that looks to avoid exceptions rather than manage them.

The demand sensing features are impressive, but the forecasting abilities are only as good as how they can support the deployment activities in the supply chain. Very specifically, the ability to generate an accurate daily forecast, auto-correct the short-term forecast and incorporate multiple signals of uncorrelated data is a game changer.  But it’s just the beginning.  These capabilities will seem quaint in a few years, but they’re revolutionary today. 

 

SAP’s IBP solution makes these capabilities palatable and accessible allowing any supply chain to reduce exceptions, increase and reduce costs.  These features enabled by machine learning are coming faster than ever, and demand sensing is a great way to introduce your supply chain to the next generation of capabilities today.

Share the knowledge

Share on linkedin
Share on twitter
Share on facebook
Share on email

Fresh supply chain knowledge delivered straight to your inbox

By signing up you agree to receive our quarterly newsletter