A pharmaceutical company produced a compound using three ingredients. One of them, Substance A, had a tendency to stick to the grinding machine, causing shut-downs and costing money for maintenance and clean-up. The producer decided if they ground Substance A finer, they could prevent it from sticking to the machine. The change was within the process specifications, so they made it and their headaches then went away. Shortly afterward, across the street, at their sister facility that blends the three substances together, they opened the barrels of Substance A and found it caked solid--only removable with a hammer and chisel. The finer grind solved the problem of sticking, but led to a new problem—caking.
Recognizing Solution-Caused Problems
Many of the issues our clients ask us to facilitate can be characterized as solution-caused problems. That is, the client has a problem, they find the cause, they put a corrective action in place to fix the problem, and all of a sudden, they have a different problem, and often a much bigger one.
Like the caking problem, the following examples, revised and disguised to protect client confidentiality, are typical solution-caused problems:
The switch: A chewable tablet failed a hardness test and became too hard after 12 months of aging. Speculation in the company’s lab focused on how the winter weather’s low-humidity might have caused excessive hardness. But the weather was no drier than previous winters, begging the question “Why did the hardness begin when it did?” It was finally discovered that, unknown to the drug manufacturer, a supplier had changed the starch content of one of the tablet’s excipients — with no thought to the effect on the final product — increasing the starch content by more than 25 percent. That increase, combined with the low humidity, caused the hardness failure.
The fix: When ‘black specks’ appeared in an ingredient, the manufacturer’s analysis identified them to be small pieces of shredded gasket material. These findings were sent to the ingredient supplier, who quickly responded that they had corrected the problem by inserting a 704 stainless steel mesh filter to separate out the black specks. The black specks disappeared, and everyone was happy. But a month later, the client began noticing ‘shiny specks’ in the same ingredient from the same supplier. When analyzed, these shiny specks turned out to be 704 stainless steel.
The improvement: A billion-dollar-a-year drug failed appearance tests for color, suddenly putting patient confidence, company revenues and shareholder value at risk. This top-selling tablet was supposed to be white, but instead was coming out dark yellow. While safety and efficacy were determined to be uncompromised, the pills just didn’t look right, especially to patients who had previously been using them. As a result, the manufacturer suspended production for more than six months. The cause was traced to a supplier who unilaterally decided to remove a substance with some potentially undesirable effects from part of the blend. The supplier believed he was being a good corporate citizen, but unknown to him, the ingredient they swapped out had played a key role in stabilizing the color; without it, the hue varied unacceptably.
Types of Solution-Caused Problems
Solution-caused problems are expensive, disruptive, and much more common than may be suspected. When discovered and analyzed, they lead to embarrassment, some finger-pointing, and a lot of head-wagging. If we deconstruct them, we can see that they come in different types and each should be approached differently.
The ‘stainless steel specks’ problem is an instance of failing to find the root cause in the first place, and of adopting an interim action instead of a permanent corrective one. Think of it as a ‘containment-caused problem’. The unasked question up front is clearly, “Why was the gasket material getting into the blend in the first place, and how could this be prevented?” Perhaps the gaskets have changed in size or shape or composition; perhaps the recommended replacement period has been exceeded; perhaps some change in the process – in speed, temperature, or pressure – has subjected the gaskets to unexpected wear.
Problem - Relocation Problems
In the caking problem, the change did solve the local problem of sticking, but caused a down-stream problem of caking. Again, failure to consider potential problems with a corrective action led to a new problem. Perhaps some potential problems were considered within the grinding process, but the range did not extend to the mixing process; the organization would have immediately vetoed the change if they had been aware of the effects it would have on other manufacturing areas.
We see this often in process re-engineering or Six Sigma, where it goes by the name of ‘sub-optimization’. In the problem-solving sphere, it looks like the old carnival game of “Whack-a-Mole” – each time you hit a symptomatic mole, it pops up again somewhere else. What is needed here is not more or harder whacking, but rather a better, more precise hammer that gets past suppressing the symptoms to attacking the root cause.
The ‘hardness failure’ and ‘appearance failure’ cases might better be labeled as ‘opportunity-caused problems’. Someone changed a variable, for what they thought was a positive reason, without thinking it would affect the final product. Once again, no one asked what might go wrong if they implemented the change; after all, they were trying to improve the process / product, not degrade it. However, when taking advantage of an opportunity, you are taking an action, and actions can have unintended consequences. You are also making a change, and causes come from changes.
All solution-caused problems are compounded by communication problems. As the prison guard in the classic movie, Cool Hand Luke, was fond of saying, “What we have here is a failure to communicate.” In all of the cases cited above, someone failed to communicate a change that could affect the process one or more steps down-stream, blindsiding those who worked where the problem manifested itself.
Anytime you introduce a change into a process, you are potentially introducing variation. It does not matter whether the change comes from trying to resolve a deviation from expected performance, or from trying to optimize a process. Changes cause problems, and change is change--it needs to be adequately analyzed and managed. Particularly today, when ever-more Lean systems are ever-more tightly connected, second-order cause-of-the-cause and effect-of-the-effect ripples need to be carefully considered.
Avoiding Solution-Caused Problems
Solution-caused problems are surprisingly pervasive, but they can be avoided. There are three elements required to minimize the occurrence of solution-caused problems and, if a problem does occur, to reduce its impact without creating more problems.
1. An Analytical Approach – Asking the question, ‘What could go wrong?’ is a start, but solely asking the question, and even listing a few potential problems, will not by itself minimize the chance of something going wrong. Our experience tells us that you have to be quite detailed about the potential problems, specific enough to be able to hypothesize some likely causes for each potential problem.
2. A Change Management System That Builds in An Analytical Approach - In the heat of the moment, such as when trying to get a costly line back up and running again, people may skip steps in order to speed up the process. One of the first steps skipped is asking, “What might go wrong?” As a result, we find that building a Potential Problem Analysis step into the approach is required to channel behavior. It may sound cynical to say, but most people are not rational unless they have to be, and will tend to avoid painstaking analysis if they can.
3. A Learning Culture - Finally, to apply Potential Problem Analysis, the company must build a culture that accepts the fact that unanticipated problems will occur, and believes that it is better to consider them in advance than try to react to them after they occur.
One subtle cause of a non-learning culture is in the human performance systems which companies develop to reward employees. When it comes to problems and potential problems, there is a built-in structural asymmetry. It is easy to see if someone has solved a problem – all you have to do is look to see if the product or the processes associated with it are up to spec again. On the other hand, it is close to impossible to ascertain whether someone has successfully prevented a potential problem from occurring. The only aspect available to examine is the fact of non-occurrence, which can be explained by assuming there never was a potential problem in the first place, or that some other unplanned event prevented it from occurring. In short, you cannot prove that your preventive action minimized the probability of the problem occurring, or that your contingent action minimized the effects, thus leading companies to recognize employees who solve problems, but not those who prevent them.
Solution-caused problems are all too common and are an indication of an incomplete approach to resolving issues. Companies can quickly earn back multiples of the time and money they invest by installing the skills needed to attack problems, systems to track them, and a mind-set that values preventing them.