The Infosys global supply chain management blog enables leaner supply chains through process and IT related interventions. Discuss the latest trends and solutions across the supply chain management landscape.

« A big fat geek wedding- JDA weds i2 | Main | Nostradamus, 2012 and Cloud Computing!!! »

Demand sensing- is it a solution to capacity planning or near term deployment planning problem?

I came across an interesting blog on Kinaxis website which talks about the importance of demand sensing especially during the post recession recovery period (Is the supply chain finally being recognized by the mainstream as a strategic capability of a company?)  It talks about how Nokia could have avoided Q3 loss of $832 m, had it utilized demand sensing capabilities to sense demand upturn and accordingly adjust capacity well in advance. Using demand sensing, they could have (theoretically) looked at the future customer orders (or other leading indicators) and sensed that they are hitting them faster than their forecasted pace and used that information to identify the future capacity issues. However, I feel, even with a good demand sensing system in place, Nokia would have still faced the problem. To elaborate further, let me first explain how demand sensing system works.

There are three essential things that are required to implement demand sensing capabilities. For simplicity, let’s just call them indicators of future, past and the present.

SKU level forecast (without the benefit of demand sensing) serves as an indicator of the future. Demand sensing system uses this forecast as a reference while applying its magic to make it more accurate at a less time granular level. This input forecast is generally obtained from the demand planning system and can be at a higher level of aggregation (typically weekly or monthly level of aggregation).

Shipment history serves as an indicator of the past. This is used by the demand sensing system to develop a mathematical model of the past sales pattern. The system uses this information to break down the input forecast into daily level forecast.

Open customer orders serve as the indicator of present market activities. This is one of the main inputs for the system which helps it sense the pace of the product sale, and use that information to adjust the daily level forecast generated by the demand sensing system.

What comes out at the end is a daily level forecast at the SKU DC level that benefits from the current event information contained in the customer orders. This granular forecast is then fed over to the APS systems to drive the deployment and production mix decisions.

Assuming it would take at least two months to increase capacity; demand sensing needs to be able to predict the shortage situation at least two months in advance. Since open customer orders is a key factor that drives the accuracy of the demand sensing forecast, there should be open orders at least two months out and beyond in order to obtain accuracy that is better than their existing demand planning accuracy. Given the nature of the product Nokia cells (pun intended), I am guessing their order visibility will not be beyond a week or two, which means, forecast accuracy outside of the two weeks should be similar to that of their demand planning system’s aggregate forecast. Based on the above statements, I would like to conclude that demand sensing system alone could not have helped Nokia predict the sales upturn.

Here is my general question to the experts out there- is demand sensing really a solution for capacity planning? Or is it a tool to enable better deployment and production mix planning? Your comments are welcome.


My question is why selling handset business order visibility is not more than 2 weeks, vs 2 months for the rest of industries.
And if 2 month is good enough for capacity planning for rest of the folks, why would 2 weeks not be sufficient for Nokia, given the fact that they know they can't look beyond 2 weeks.
If your line of thinking is correct, then handset industry can't predict demand, which should not be the case.

Punit, I am not saying wireless handset industry has order visibility of 2 weeks while others have 2 months. I come from CPG industry where order visibility curve peaks at 3 days yet they do demand forecasting. The point I am trying to drive here is that Demand Sensing is effective only within the period of order visibility, beyond that the traditional demand forecasting should be trusted more. Having said that I have no idea what level of order visibility Nokia or related companies typically have, I have just tried to make an intelligent guess. If anyone has insights on the actual numbers, I would be happy to listen.

Post a comment

(If you haven't left a comment here before, you may need to be approved by the site owner before your comment will appear. Until then, it won't appear on the entry. Thanks for waiting.)

Please key in the two words you see in the box to validate your identity as an authentic user and reduce spam.

Subscribe to this blog's feed

Follow us on

Blogger Profiles

Infosys on Twitter