The motivation behind this post is a personal experience I had with my washing machine. I (like any typical Amazon user), was looking for a washing machine in the price range I was ready to pay for, then checked the specifications, and finally the user reviews. Based on these criteria I made the decision and bought a machine of a well-known manufacturer.
The price was below my budget which was great, the specs for my needs were good, and best of all it had five stars rating and very positive reviews, this was in the second half of 2012.
What the figure below does not show, is that some of the one star ratings prior to 2015 were actually positive ratings that were edited later after the machine went broken or the user changed her mind).
For many customers, the machine was by 2015 (the machine was listed on Amazon in Nov. 2011) already irreparably broken.
By irreparable I mean the cost for buying and changing the broken part is similar or more expensive than buying a new machine.
I was lucky, it lasted for me about 7 years before the ball-bearing broke. Changing it means buying a whole new tub (the cylinder where you put the clothes which rotates plus the stationary plastic cover for more than 300€, while the actual broken part should not cost more that 25€.
Older models had a tub that was not glued or sealed but its two parts were screwed together, changing to a sealed tub seems to be a strategy to plan the obsolescence of the machines, which many manufacturers are doing.
Does the manufacturer know approximately when a product would stop working? I would say yes. What could explain the sudden drop in ratings around beginning of 2015? If we would to fit a function to the ratings, it would look like a step function.
Because of the glued tub, if any part inside 0320 breaks, one needs to replace the whole of 0320.
Washing machines are nowadays a commodity product which means, you can easily buy one and you expected it to work. In such a market, manufacturers try to differentiate their products and rely heavily on marketing.
While researching for this post, I noticed that academic literature on washing machines design and modeling that are funded by manufacturers concentrate on reducing the vibrations of the machine. While this does not directly improve the functionality of the machine, it gives a good impression for the user and might even prolong the life of the machine.
Most eCommerce platforms including Amazon, do not show out-of-stock products, so the manufacturers can accurately calculate the product release cycle safely, and the users/customers would only see mainly positive ratings.
At University I attended a course on Modeling Reliability of Systems, which was partly concerned with building stochastic models of reliability for products with multiple components that can fail with different probabilities.
Well-known bearing manufacturers (like SKF and NTN) provide both generic and manufacturer-specific formulas and calculators for calculating the life rating of a bearing as a function of load or under assumption of working under the specified load for that bearing.
So, all these tools are well established techniques that we can safely assume that machines manufacturers can and are using to model the life of their products.
According to NTN’s document, one of their bearings with a home appliance application can have a 90% reliability with a life rating of 4000 hours (assuming a given rpm). If we assume that a household uses the washing machine twice a week, and each wash takes two hours, we end up with 19 years with 0.9 probability. So what went wrong here? Why the bearing was broken for so many people after less than four years.
It is hard to believe that any manufacturer of mid- and low-end machines would like to see his product working for long years. Some people defend the planned obsolescence as a way to support research and development of new devices. But to be honest we do not see any revolutionary or even interesting new feature in newly released home appliances.
Back to the title of the post, for many products, the customer ratings when the product is newly released draw a picture of the first impression of the product. In the case of a washing machine, it is mostly how quite the machine is.
As the figure above shows there seems to be a strange relationship between unlisting date (after that date, Amazon users would not see the product when searching for such a product) and the point when the ratings went from 5 to 1 stars!
*https://www.amazon.de/Bosch-WAQ28321-Waschmaschine-VarioPerfect-Mengenautomatik/dp/B006C0KWLO/