Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for
Search instead for
Did you mean:
Our World Statistics Day conversations have been a great reminder of how much statistics can inform our lives. Do you have an example of how statistics has made a difference in your life? Share your story with the Community!
Setting Goals for Product Return rates based on History Data
Feb 11, 2019 6:52 AM(2678 views)
We track monthly return rates of each of our products. Always have a hard time setting yearly return goals. I have been working on setting if you Guardbands that we plan to use for the next year goals. Know this process has issues, using a Monte-Carlo Ditribution. Would like to ask about other methods, have tried the JMP Proflier, and Control Charts.
Data Sets typically look like this but over a 24 month time frame.
1.25%, 2.12%, .28%, 3.1% 2.2% etc....
One of the sources of varability is that each product and product family is so differnt.
My experience is that a 'goal' for any process typically don't come from the process data itself...but from some external source...usually management. Any computational algorithm you might use to establish say control limits, or say, for a time series model, prediction limits, will provide a number which can in turn be converted into a goal. It's just that my past experience in business process monitoring is that 'goals' are always established by management...not the process and it's inherent performance. Can you provide some insight around why these 'goals' are needed and to what actions or decisions the process performance wrt to the goal will drive?
Very good input. Agree in the past we have tried basing the goals on product complexity (1st gen vs. 3rd gen), reliability indicators, and past performance to set Product Goals for Returns. Also have worked in Warrenty Cost Reserve, and what inpact incermental product changes has made over a specific family of products over the life of the product. Always turned out to be a mess.
Just using historical data is an approach to hopefuly to ID a target goal or some type of "guardband", that if we exceed will drive action from Engineering and the Manuf. team to address the excursion. Plan to make goal adjustment once a product leave the "ramp" phase and stablizes. Bassed on our breath of products, we need a screening factor to flag issues, the indicators will also be used for performance metric to the different Product teams.
So it sounds like you are really trying to establish limits of process performance metrics that will incite/initiate action. Sure sounds like a process monitoring approach (control charts?) to me...from there it's putting a process monitoring regimen in place that is a combination of data collection, analysis, conversation, decisions (and yes, 'doing nothing' is a viable decision option), and ongoing validation for reasonablness. For a view of 'what's possible in JMP' I can suggest a few on demand webinars that might help you: