# Forecasting of Product Returns and Warranty Costs

L

#### lord85

Hi Everyone,

We are a pump products manufacturer and we are planning to release new products to the market. I will put some targets for product' performance. I want to forecast possible product returns and warranty costs after some months. Are there anybody who can help to calculate this or suggest a formula? Which forecast models (weibull, waterfall, weighted average etc.) do you use?

in advance.

#### Bev D

##### Heretical Statistician
Leader
Super Moderator
straightforward return rates are best. Weibull can put some more statistical precision to the estimate. but the first place to start is: what data do you have so far? have done stress testing? do you have initial life data? The formula you use is dependent on the data you have

L

#### lord85

Dear Bev D,

We are at the start of the project and planning to release the product at the beginning of the 2016. We have done some calculations and tests about the product life but the real life doesn't match generally the calculations because of the production inputs. The base formula for product returns will be [(total failure quantity after sale / total sold product or total produced product) * 1000000] (ppm) and be calculated for every month. Also it should take into account previous produced products for 6 months. I mean the performance review will be for example from 6th month of the year 2015 to 6th month of the year 2016 for whole year of 2016. But the forecast should say to me the possible rates for next period. The system is alive and changeable.

#### Steve Prevette

##### Deming Disciple
Leader
Super Moderator
Yes, some form of reliability analysis ought to be able to predict failure rates from results of your intial product testing data and/or specifications , in conjunction with project sales figures.

Then, once you start gathering actual returns and warranty costs, some form of statistical analysis (such as SPC) is in order to see if the returns and warranty costs (after some form of start up ramp) reach stability and actual predicatability.

Yes, I'd be willing to by PM assist, but yes, would need access to your existing data.

#### charanjit singh

##### Involved In Discussions
I have a few questions for you on this subject:

1. Is this pump a totally a new design?
2. Is it a modified/improved version of some of your existing products? Have
you at least theoretically evaluated the effect of the changes on the
performance/useful-life of the modified model?
3. If it is modified version, do you have performance/field returns/warranty
failure data of the old version?.
4. Is this pump meant for a new application? If so, have you evaluated some
prototype models under the anticipated application/environment? How
does the performance of prototypes compare with the design
specifications?
5. What is your field-experience (customer complaints, warranty
replacements etc.) with various products that you manufacture
and supply to market?

There could be some more questions to consider, but for the present these. I think these are of primary consideration.

Hope this helps.