Checklists at Work: Why They Prevent Errors and What Good Ones Look Like

Why do experienced people make simple mistakes?

Whoever has done a workflow a hundred times no longer thinks about it. That is the advantage of experience: it makes you fast. It is also its risk. Under time pressure, after an interruption, or at the end of a long day, a step drops out that you actually know. The quote goes out without the agreed payment terms, the machine is released without a visual check.

Checklists are the oldest remedy, and they have a dubious reputation. Many teams know them as forms that are ticked off after the work is done so that the file is complete. This leads to the guiding question of this article: When do checklists actually prevent errors, and how must a checklist be built so that it works in daily practice rather than just being ticked off?

To answer it, we first clarify which kind of error a checklist can prevent at all. Then we look at the best-known application, the surgical checklist, with its successes and its limits. At the end there are rules for good checklists, an example from a small business, and a step for this week.

A checklist does not protect you from what you do not know, but from what you know and forget anyway.

Ignorance or inattention: which errors does a checklist prevent?

The surgeon and author Atul Gawande popularized a distinction in his book “The Checklist Manifesto” (2009) that serves as the thinking model for this article. Errors have two sources. Errors of ignorance arise because the necessary knowledge is missing, for the individual or in general. Errors of ineptitude arise even though the knowledge exists: it is not applied correctly or completely at the decisive moment.

In simpler times, the first source dominates. In a company with experienced people, established workflows, and many simultaneous tasks, the second dominates. A checklist works only against this second source. It is, as Facey and colleagues (2024) summarize Gawande, a “cognitive net” meant to catch lapses in attention, memory, and thoroughness. Against missing knowledge, by contrast, onboarding, a good work instruction, or an expert helps.

This distinction explains why long checklists fail. Whoever tries to write the entire knowledge of a workflow into a checklist builds a work instruction that nobody reads at the moment of work. How to write work instructions that get read is described in the article on writing work instructions. The checklist is the smaller tool: it contains only what experienced people forget.

What does the surgical checklist show?

The best-known study of checklists comes from surgery. The World Health Organization developed a 19-item checklist that is gone through at three pause points: before anesthesia, before the incision, and before the patient leaves the operating room. Haynes, Weiser, and colleagues (2009) tested it in eight hospitals worldwide. According to the summary by Facey and colleagues (2024), studies by this group reported a reduction in mortality and complications of about one third, and another study a 37 percent decrease in postoperative mortality.

The results led authorities to mandate the checklist. In the Canadian province of Ontario, reporting was tied to hospital reimbursement in 2010 (Facey et al. 2024). After that, the findings became more mixed. Urbach and colleagues found little or no significant effect on mortality from the province-wide introduction in Ontario. A systematic review found no reduction in mortality but fewer complications. And a comparison of many studies found differences of more than 30 percentage points between documented and observed completion (all summarized in Facey et al. 2024).

How large this gap can be is shown by a study by Rennie and colleagues (2026). They analyzed video recordings of 212 procedures in one operating room. The checklist was started in 87.7 percent of cases. According to documentation, on average 95.5 percent of the items were completed; 46.8 percent were observed. Experienced surgeons started and completed the checklist less often than less experienced ones.

Effective tool or ritual? Two readings

There are two readings of these findings. The first: checklists work, but only if they are really done. The mixed results of later studies then say nothing about the tool but about its implementation. The second reading is more fundamental. Facey, Baxter, Hammond Mobilio, Moulton, and Paradis (2024) observed the use of the checklist in a Canadian teaching hospital and describe it as a ritual: a fixed sequence of actions that serves to visibly comply with requirements. The nurses documented reliably; the surgeons performed the list in an improvised, abbreviated way. The link to the actual goal, patient safety, became, as the authors write, opaque.

The two readings contradict each other less than it seems. The first says what a checklist can achieve. The second says how it loses that effect: when it is imposed from outside, kept as proof, and not taken seriously by the most experienced. The studies come from hospitals, that is, from a high-risk environment with strong oversight. For a small company, the finding is still transferable, and it is rather encouraging. A company needs no reporting obligation that tempts people to tick boxes. It can build the checklist so that it serves the work.

What does a good checklist look like?

Gawande (2009) distinguishes two types. With a read-do list, you work through the items in order, like a recipe. It suits rare workflows. With a do-confirm list, you work from memory and pause at a fixed point to check whether everything important is done. It suits routine and is almost always the right choice in a small business.

From Gawande’s recommendations and the findings on rituals, a four-question test can be derived. It serves as a tool for checking any existing checklist.

• Pause point: Is there a fixed moment at which the list is gone through, such as before sending, before handover, before starting?
• Brevity: Does the list have at most nine items and take about a minute? Gawande generally recommends five to nine items.
• Selection: Does it contain only items that actually get forgotten and whose omission is expensive, and nothing everyone does anyway?
• Together: Is it, where possible, gone through by two people, one reading and one confirming, rather than ticked off alone?

The third question is the hardest. The temptation is great to include everything that has ever gone wrong. But a list with 25 items is not gone through; it is skimmed. The question for each item is: has this been forgotten at least once in the last twelve months, and would it have cost money, time, or a customer?

Example: checklist before sending a quote

A company with a small sales team has noticed that quotes keep going out with errors everyone knows about. The following do-confirm list is gone through immediately before clicking “send.” The items are an example and must be adapted to your own company.

The second column is notable. Every item has a reason that comes from a real mistake. That makes the list credible, even to the most experienced. How the entire quoting workflow is standardized with templates and approvals is described in the article on how to standardize the quoting process. The checklist is the last pause point there.

Other typical pause points in small companies are the handover of an order to production or the project team, a new employee’s first day, acceptance at the customer’s site, and the monthly close. For the handover from sales to the project team, there is a dedicated list in the article on the sales-to-project handover.

How do you introduce checklists without creating a ritual?

1. Start with a mistake

   Take a mistake that has happened several times in recent months and annoyed everyone. The list is the answer to it, not a project.

2. Write it with the people who do the work

   The list is created in 30 minutes with the people who perform the workflow. They know what really gets forgotten.

3. Define the pause point

   Determine the one moment at which the list is used and put it exactly there: in the template, at the workstation, in the tool.

4. Test for two weeks

   Everyone uses the list. At the end you ask: which item caught something, which never did?

5. Shorten

   Items that never caught anything are removed. Missing items are added as long as the nine-item rule holds.

6. Observe instead of counting

   Occasionally check whether the list is used at the moment of work, not whether it is complete afterward.

The last step is the lesson from the video study by Rennie and colleagues (2026). Documentation says little about whether a checklist is actually used. A look over the shoulder the next time a quote is sent says more.

If the same mistake recurs despite a checklist, the cause often lies in the workflow and not with the person. Why that is so is described in the essay on errors as process signals. And how a team investigates a mistake together without looking for culprits is shown in the article on learning from bad decisions. A checklist can be used long term only in an environment where a forgotten item may be raised openly.

Conclusion: when checklists prevent errors

We set out to ask
checklists actually prevent errors and how they must be built to work in daily practice. The answer starts with the type of error: checklists help against forgetting what is known, not against missing knowledge. They work when they are short, are used at a fixed pause point, contain only the items that really get forgotten, and are taken seriously by the team.

As shown, the first studies from the operating room demonstrate clear improvements, while later studies of mandated introduction show weaker effects and a large gap between ticks and reality. This does not lead to a verdict against checklists but against their use as proof. The sober observation is: a checklist works exactly as long as the team sees it as help and not as control.

How to get started

1. Ask your team: which mistake that we all know about has happened more than once in the last three months?
2. Identify the pause point at which it could have been prevented.
3. With the people who do the work, write a do-confirm list of at most nine items in 30 minutes, each with a reason.
4. Place the list exactly at the pause point and test it for two weeks.
5. Then check every existing checklist in the company against the four-question test and shorten or remove whatever fails it.

If you document your workflows anyway, you can add the list as a resource on the process page, as the one-page method provides.