Yesterday I blogged about Refrigerator warranty myths and the bathtub curve. I thought I would add a little more meat around the warranty prediction question, especially for those new business startups that need a little help.
Of the many people I get to meet in different circles who pitch their great ideas to me with total confidence, I often break their hearts when I ask them what their returns plan is. They look at me puzzled and say, “Well I don’t expect the product to break…”
Unfortunately, many banks or private investors want to understand your risks and one of the key ones is about product returns. This is not necessarily implying that your product will fail, but the notion that your customers will return product for various reasons that will erode your revenue and add additional costs such as returns (RMA) processing.
In this blog, I won’t go into reasons why customers return products but keep in mind that products are returned for being defective, not what the customer wanted, and even buyer’s remorse fits into this.
What I want to talk about here is how to predict your product returns count and how to set a reasonable warranty period based on a warranty prediction analysis.
As more and more people become independent and start small companies, I also recognize that there are many that do not incorporate reliability engineering into their designs or products. The reason is simple: more and more entrepreneurs are OEM’ing products that are increasingly available and simply rebranding and repacking many of them after getting bulk pricing from sites like alibaba.com or even ebay.
Minitab’s Warranty Prediction Tool
Many warranty prediction software tools are available but my favourite is in Minitab. Minitab offers you the triangular matrix warranty analysis which allows you to enter the number of units shipped and the number of units returned for a predetermined future period.
This of course, is assuming you have historical data for the future analysis. For example, how do I figure out what my warranty prediction will be if I have never shipped any product? You can contact me (firstname.lastname@example.org) for help at guesstimating, or you can set a pre-determined rate such as 20% failure in the first year. Setting a 20% failure in the first year is high as most product designs strive to keep this number at 2% or lower, but by setting a 20% failure rate, you are positioning your product for the consumer market. 20% returns in a consumer market is not unheard of and probably 60-80% of those returns are not defect-related, but I digress.
Let’s suppose that you have some historical returns data. You have shipped 100 units per month for a total of 1200 units shipped in one year with 73 returns. Minitab allows you to enter the number of returns per month, and using typically a Weibull distribution, will offer you a warranty model for future periods. A key note here is that this will be for future periods.
So whatever failure rates you have for the first twelve months (73 in this example), the future period will highlight the number of failures for the next twelve months. If you intend to offer a 24 month warranty, then you must factor in the year 1 costs and the future year 2 costs.
In the example below, it is estimated that by the end of the 24 month period (future 12 month period) 263 returns would be expected on 2400 units shipped. Let’s break this down: 1200 units are shipped for each year with a total of 2400 units shipped in two years. 73 units are returned in year 1 and 190 returns are expected in year 2. 263 units returned on 2400 units shipped would suggest 10.9% return rate over two years. Not bad.
After analyzing the actual returns, failure experts explore whether there are required design improvements or documentation improvements to better demonstrate how the product should work.