“The language we use matters.” Wayne Nelson
How we talk about reliability does matter. It sets expectations and influences decisions. We talk about reliability as it is important to our customers and our business. The successful operation of the equipment in a plant permits the production of products. The successful operation of the product satisfies the needs and expectations of our customers.
The ability to translate our customer and business expectations into words relies on the language we use. It has to be clear.
A customer may say, “It should never fail.” This is clear. While we would like to provide them a product that doesn’t have a chance of failure, that is not possible.
Most customers are willing to accept a small risk that a product will fail. Most customers understand the tradeoff between features, price, and durability. It is finding the right balance and setting the reliability targets appropriately.
‘Good enough’ reliability
Once a small military communication device manufacturer received a complaint that their device failed too often. A little data analysis found the device failure rate was lower than the contract requirement, thus met their stated expectation. The customer followed up with the notion that the device failed more often than anything else on the aircraft.
It wasn’t the original expectation, it was the similar equipment that made the failure rate unacceptable.
Products and equipment exist to serve some purpose, and when they do not do so, they are compared to other similar devices along with the expectations held by the customer. It is this notion allows us to create products with ‘good enough’ reliability.
It’s like the joke about two men on safari alone and they notice a lioness stalking them. One ties his shoes as the other prepares to flee and stops to ask, “what are you doing? You can’t outrun a lion.” And, the guy with his shoes on says, “Don’t have to, just have to outrun you.”
A simple way to set a reliability target is to evaluate your product against the previous model and your competitor’s models. It just has to be “as good or better” than the others.
Is this clear? In what way, under what stresses, and according to what criteria do we make this judgment? Our customers will make this judgment in real-time. We often do not have the luxury of real time testing under the full range of use conditions. And, how do we make decisions that lead to the creation of a reliable product?
I once was told by a medical device general manager (that didn’t last in the role very long) that “all customers complain, it’s not a big deal.”
They had a 50% first-year failure rate and were losing market share quickly.
In business, we want to meet or exceed customer expectations concerning reliability. It helps the customer to appreciate the brand, buy more products, and recommend a purchase of our product to others.
There also is the cost of failure, like the cost of quality, it is often seen as warranty. The cost of product failures can take many forms:
- Warranty Costs
- Lost Sales
- Lost engineering time to resolve field failures
- Product scrap if unable to repair produced units
These are just a few examples.
Cost of fixing field failures
One of the biggest and often hidden is the cost of engineering time used to resolve field failures. In my experience, this is often 25% of the engineering talent of an organization. How about just getting the product right before shipping? The savings is not just warranty costs.
One way a business may translate the expectation of ‘not too many failures’ is by setting a warranty target. Another is to estimate the customer’s useful life expectation. In both cases, we have the ability to set both a duration and probability of success target.
For example, if a business would like to spend less than 2% of net revenue on warranty. Then a little math with expected sales volume and cost per failure will establish the target probability.
You could include the cost of engineering time or other costs of failure components to influence the target value.
It’s not hard.
Have a clear target
The first time an organization does this exercise and compares the target to actual results, they determine the magnitude of the difference. If on target, great. Often though there is significant room to improve.
This exercise connecting the business objectives of profit, costs, and resources to the reliability goal allows the entire team to assess the importance of achieving the reliability targets. It provides a means to trade off the many priorities in each decision during the product lifecycle. The target just has to be visible and connected.
The target just has to be visible and connected.
This is just an outline. There is work to be done if your reliability targets are not clear and connected to your business and customer expectations.
Let’s summarize with a simple example. Let’s say we are designing a remote control for the companies entertainment system. Our customers are the people that buy the larger system and they receive in the box the remote.
If asked a customer may expect the remote to last as long as the entertainment system. Which makes sense. A little market research may suggest the system should last 10 years. And, the common warranty duration in the industry is one year. We have two durations for our reliability target.
The probability of success values might come from the business objectives concerning profitability. Let’s say the cost of replacing a failed remote is two and a half times its retail price and while rare it is possible the entire system is returned based on a faulty remote at a cost of 1,000 times the cost of the remote.
After a little math based on costs and volumes, we estimate we would like less than 0.5% remote control device failures over the first year of operation.
Reliability and customer satisfaction
The ten-year useful life is roughly connected to the customer satisfaction.
This is related to brand loyalty and probability of recommending the brand to others. And, the converse.
This is hard to quantify, yet the marketing and sales team likely have a figure for the cost of obtaining a new customer versus the cost of reselling to a loyal customer.
Also, consider the companies policy on failed remotes that are outside the warranty period. It may make sense to replace the remote for free or reduced price to maintain customer satisfaction with the larger system.
Again, with a little math, we may conclude that we would benefit if less than 5% of the units failed during the first ten years of use.
Based on the logic and using the example we can clearly state the following reliability target:
The XYZ model entertainment system remote control device will provide control functionality (see product requirement document xyz for specifications) for the home entertainment system in North American homes with greater than 99.5% chance of successful operation over the first year (warranty period) of use, and 95% chance of successful operation over the first ten years of use.
We want 99.5% reliability in the first year and 95% reliability over ten years.
That is clear, right?