When Standards Do Harm: Designing Room for Exceptions Deliberately

Why good standards sometimes produce bad work

This website often recommends standardizing: templates for quotes, process pages, checklists, work instructions. There are good reasons for that. Recurring work that everyone does differently costs time and produces errors. But there is also the other experience. A customer with an unusual request gets the standard answer. An employee knows that the prescribed way is wrong in this case and takes it anyway, because deviating causes trouble. Or everyone deviates, and nobody says so.

This essay therefore asks: When do standards do harm, and how does a company create room for exceptions without losing the benefits of standards? To answer it, we first clarify what a standard is, then introduce a thinking model from organizational research, name the situations in which standards do harm, weigh two perspectives, and finally arrive at a rule for exceptions.

A standard nobody may deviate from is not followed but kept quiet about.

What is a standard, and how does it differ from a regulation?

A standard here is the defined, currently best-known way of doing a recurring piece of work. The guide to standardizing the quoting process puts it in one sentence: a standard is a model that allows deviation. A regulation, by contrast, is a rule that may not be deviated from, such as a safety step or a legal obligation. Discretion is the area in which the person doing the work decides for themselves, and an exception is a deliberate deviation from the standard in an individual case.

Most problems arise when these terms blur. If every standard is treated like a regulation, discretion disappears, and with it the judgment of the people who know the work best. If every regulation is treated like a standard, safety disappears at the few points where it really matters. The guide to writing work instructions solves this by separating a few binding, justified steps from the rest.

The thinking model: enabling and coercive standards

Paul Adler and Bryan Borys (1996) wanted to explain an old contradiction. One part of the research found that written rules alienate employees. Another part found that they help them do good work. The authors resolved this by looking at rules as a technology, much like a device. A copier can be designed to need the user as little as possible and to call the technician at every paper jam. It can also be designed so that the user sees what is happening and clears the jam themselves.

Accordingly, they distinguish coercive formalization, which is meant to secure effort and compliance and treats every deviation as suspect, from enabling formalization, which helps employees master their tasks better. Four features separate the two.

As an example of enabling standards, the authors describe the NUMMI car plant in California, a joint venture of Toyota and General Motors. There, workers and supervisors defined the work methods together and in great detail. A deviation was treated as a signal: either of additional training needs or of a standard that needs revising. Remarkably, the rules there were more detailed than in many other plants. The difference lay not in the number of rules but in their purpose.

In which situations do standards do harm?

From the model and further findings, four situations can be derived in which a standard does more harm than good. They are not a complete list, but they cover the most common cases in a small company.

1. The work is uncertain or ambiguous. Zelt, Recker, Schmiedel, and vom Brocke (2019) proposed a theory according to which means such as documentation, standardization, and monitoring must fit the nature of the process. The more uncertain and ambiguous a process is, the more coordination it needs and the less a fixed procedure helps. An invoice can be standardized; a custom-made product only in its framework.
2. The exception is the rule. If a standard covers only a small share of the actual cases, everyone works in the exception. Then the standard describes the work as imagined, not the work as done, to use the terms of safety research.
3. Deviations may not be reported. Hale and Borys (2013) show in their review that rules written from above and enforced through control do not prevent deviations but make them invisible. The standard then becomes outdated without anyone noticing.
4. The standard replaces judgment instead of supporting it. Whoever only works through text modules in a customer conversation stops listening. Hanahara (2026) describes that standardization in sales held where a shared basic pattern was practiced, and not where detailed procedure manuals were handed out.

Foundation of improvement or shackle? Two perspectives

The first view comes from lean production and is often attributed to Taiichi Ohno of Toyota: without a standard there is no improvement. Only once it is clear how a job is done today can you check whether a change is better. A company without standards does not improve; it merely changes. This view is supported by the fact that the PDCA cycle described in the article on continuous improvement on a small scale ends with writing down the new standard.

The second view says: every standard freezes a solution that was good under certain conditions. When conditions change, it becomes a shackle, and whoever sticks to it acts correctly and still wrongly. Especially in jobs with much customer contact and changing tasks, standards crowd out the judgment the company depends on.

I consider both views correct, but for different kinds of standards. The first describes enabling standards: developed jointly, justified, with a path for deviations. The second describes coercive standards: imposed from above, without justification, with deviation as a violation. The obvious assumption that a company simply has too many or too few rules therefore usually misses the point. The NUMMI example shows that very detailed rules can create room for discretion if they belong to those doing the work.

The exception rule: discretion that strengthens the standard

Discretion does not arise from leaving out standards but from regulating exceptions. The tool for this is called the exception rule here. It belongs on every process page created with the one-page method and answers four questions.

1. What is binding?

   Mark the few steps that are not deviated from because safety, law, or major damage is involved. Each gets one sentence of justification.

2. Who may deviate?

   Set the limit up to which the person doing the work decides for themselves, such as up to an amount or a schedule shift, and who decides beyond it. The logic matches the article on decision paths.

3. What is reported?

   Every deviation is recorded in one sentence: what, why, with what result. Reporting has no negative consequences.

4. When does the standard change?

   If the same exception occurs more than three times in a month, check whether it belongs in the standard. That way today’s exception becomes tomorrow’s standard.

The limit for one’s own decisions is the core. Who may decide what without asking is described in the guide to setting decision paths. Something similar applies to checklists: they should contain only the items that really get forgotten and not prescribe every move, as the article on checklists in the company shows.

What follows beyond the individual case

For an owner’s attitude, an uncomfortable question follows. What are standards introduced for in your own company? If the honest answer is so that management can check whether people are working, a coercive standard emerges almost inevitably, and with it the concealed deviation. If the answer is so that the work becomes easier and safer, an enabling one is more likely. Wouters and Wilderom (2008) showed in a three-year case study in the logistics department of a medium-sized beverage manufacturer that performance measurement systems were experienced as helpful when employees had helped develop them.

The model’s limits belong here too. Adler and Borys proposed a theory, not a measurement, and their examples come from large industrial companies. Whether a particular standard has an enabling or coercive effect also depends on how it is experienced, and that can differ within the same team. And there are areas in which deviation is not a learning opportunity, such as safety shutdowns or legal obligations. There, the regulation remains a regulation.

Conclusion: when do standards do harm?

We began by asking
standards do harm and how a company creates room for exceptions without losing their benefits. Standards do harm when they are built coercively: when they prescribe a fixed sequence in uncertain processes, when the exception has long been the rule, when deviations may not be reported, and when they replace judgment. They help when they are built to enable: justified, with an overview, with a way to fix disruptions oneself, and with an exception rule that keeps deviations visible.

The question of whether a company has too many rules is therefore usually misleading. The better question is whom the rules serve. The sober observation is: a standard is only as good as the way its first exception is handled.

How to get started

• Choose a process in which people often deviate from the standard and estimate how many cases the standard really covers.
• Mark the binding steps in this process and justify each one in a sentence.
• Set the limit up to which those doing the work may decide for themselves.
• For one month, keep a simple list of exceptions with reason and result.
• After the month, check which frequent exception belongs in the standard.