cancel
Showing results for 
Search instead for 
Did you mean: 

External Data for Demand Sensing

brienne_tang
Explorer
0 Kudos

Dear All,

Do you have experience in using external data purchased from information, data and market measurement firm for Demand Sensing?

Would it help Demand Sensing to generate better forecast?

Otherwise, what is the input you would recommend.

Thanks

Hello Brienne – Great discussion topic. If the external signal is valuable, it will likely improve the forecast. The quality of the signal is more important than the number of signals.

The challenge is how do you know? You can make the change and measure forecast effectiveness, which is always a good practice.

Demand Sensing can determine a signal’s usefulness. After you add the signal to the model, you can see how it adjusts the forecast by reviewing the importance (%) and impact (Qty) key figures. These key figures are time-varying, so a signal’s usefulness will likely vary by week.

Here’s another challenge to consider. An external signal may not be precisely aligned with your demand. For example, if you increase advertising in one week, you may not see the sales lift for a couple of weeks. Demand Sensing can automatically see this pattern in the signal and adjust your forecast accordingly.

I hope this is helpful.

Accepted Solutions (0)

Answers (1)

Answers (1)

lingaiahvanam
Active Contributor
0 Kudos

Hi,

It depends on the signals which you want to input to demand sensing.

Demand sensing will consider multiple internal and external signals to predict the short-term fulfilments. When we input multiple external signals make sure that data which you feed has meaningful indications.

Customer can get the better forecast when they use open orders, shipments, and mid-term forecast along with market conditions (external signals). IBP Demand 2302 release has better usability options to view the external signals impact on short term forecast. It is easy to understand the patterns and what basis forecast generated by ML algorithm.

Additional note ML algorithms will produce better forecast when customer has longer periods of history.

Best Regards,

Lingaiah