Recurring work often depends on someone remembering to do the right thing at an inconvenient moment. A review competes with incoming requests. A record update waits until the end of a busy day. A task performed only occasionally becomes urgent because nobody noticed it was due.
In Atomic Habits, James Clear explores decisions made in advance that make later behaviour easier to sustain, including commitments and automation. A business can use these ideas to reduce dependence on memory and momentary motivation. The useful aim is a more dependable process with clear ownership, rather than the assumption that a one-off setup will make future results certain.
Understand what an advance commitment changes
A commitment device is an arrangement made now that makes a chosen future action easier to follow through on, or makes departing from it more deliberate. Booking a review with a colleague creates a different situation from privately hoping to find time for the review.
The arrangement changes what happens when priorities compete. A booked session has a time, purpose and another person’s involvement. Cancelling it requires an explicit decision. The commitment can therefore make a useful intention more visible and harder to forget.
This is not proof that the original intention should always win. New information may justify postponement or cancellation. A good business commitment makes the trade-off clear while preserving a sensible way to respond when circumstances change.
Start with commitments that are modest, transparent and within your authority. A founder can reserve time for their own planning. A team arrangement needs agreement about workloads and responsibilities. Making someone else’s work harder without their involvement is not a sound application of a personal habit technique.
Distinguish reminders, defaults and automation
A reminder draws attention to an action but leaves the action to a person. A default supplies the initial choice unless someone changes it. Automation performs a defined action when specified conditions are met. These mechanisms solve different problems and need different checks.
A reminder may be enough when work is infrequent and requires judgement. A default may help when the same sensible starting choice is repeatedly selected. Automation may suit a stable, repeatable step whose inputs and expected outputs can be defined clearly.
| Mechanism | Illustrative use | Main question to resolve |
|---|---|---|
| Reminder | Prompt the owner to review an open item | Will the prompt reach someone able to act? |
| Default | Begin a new record with an agreed status | Is that starting status appropriate for this case? |
| Advance commitment | Reserve a recurring review session | Can the participants realistically honour it? |
| Automation | Create a review task when an item reaches a defined stage | Can the trigger and result be checked reliably? |
Avoid using a stronger mechanism simply because it is available. Automatically executing an action can create more risk than prompting a person to consider it. Choose the least complicated arrangement that addresses the actual failure in the work.
Choose the decision before choosing the tool
Begin with a specific recurring problem. Perhaps a review is forgotten, information is copied inconsistently or work arrives without an assigned owner. Describe what should happen, what currently happens and why the difference matters.
Then identify the decision that would prevent the gap. A forgotten review may need a visible schedule and owner. An inconsistent record may need a clearer definition of required information. Software cannot settle those questions on behalf of the business without someone defining the intended behaviour.
Observe the existing process before automating it. The person doing the work may be making useful exceptions that never appear in the written instructions. If those judgements are omitted, the automated version may handle ordinary cases quickly while mishandling the cases that most need attention.
For a process with substantial avoidable effort, Making important work easier to start offers a useful preliminary step. Simplifying and clarifying a process can reveal that a modest change is sufficient, or make a later automation easier to specify.
Use one-off setup decisions with realistic expectations
Some decisions continue to influence work after the initial effort. An agreed template can improve consistency across many documents. A shared location can reduce repeated searching. A recurring calendar entry can give an important review a visible place in the week.
These are continuing benefits, but they still depend on use and maintenance. A template becomes less useful if it contains obsolete assumptions. A shared location creates confusion if people keep competing versions elsewhere. A calendar event becomes background noise if nobody owns its purpose.
Record what the setup is intended to achieve and who maintains it. This can be a short note within the existing process documentation. The point is to make the arrangement understandable to someone who was not present when it was created.
Also decide what would make the setup unnecessary. A recurring step may become redundant when the business changes its offering or workflow. Without a review point, a helpful decision can turn into a permanent obligation whose original purpose has disappeared.
Automate stable steps with observable outcomes
A useful candidate for automation has a recognisable trigger, suitable information and a result that someone can inspect. For example, an internal process might create a review task when a record enters an agreed stage. The business can then check whether the task was created correctly.
Write the expected behaviour in plain language before implementation. Specify the qualifying event, required inputs, action, owner and exception response. If these cannot be described clearly, more process design is needed before the work can be handed to a system.
Separate mechanical steps from judgement. Moving a record into a review queue is different from deciding whether its contents are adequate. An automated preparation step can support a human decision without pretending that the decision has already been made.
Start with a narrow scope that can be compared against real work. A small trial makes it easier to examine incorrect triggers, missed cases and unexpected consequences. Expanding a poorly understood process can multiply confusion as readily as it multiplies throughput.
Make exceptions part of the design
An exception is a case that falls outside the conditions the normal process can handle. Missing information, duplicate records or an unusual customer request may all require a different response. An exception path gives such cases somewhere useful to go.
Decide who receives the exception, what information they need and how they indicate that it is resolved. A warning with no responsible person is only a record of the problem. A task assigned to someone without the necessary authority is similarly incomplete.
Avoid designing a process that appears successful by silently ignoring difficult cases. Counts of completed actions should be considered alongside unresolved exceptions. Otherwise, an impressive activity total can hide a growing queue of work that nobody is handling.
An override should also have a clear purpose. It lets an authorised person respond to a legitimate unusual situation, with enough explanation for later review. If people repeatedly use the override for ordinary work, the main process may no longer match the business.
Keep an owner after setup
Automated work still needs a responsible person. Process ownership means being accountable for the process’s intended outcome, including checking whether it remains suitable and coordinating changes when it does not.
Ownership does not require watching every action. It does require knowing how to inspect performance, where failures appear and who can correct them. The owner should understand the process well enough to distinguish an isolated exception from a recurring design problem.
Decide how work continues if the automation is unavailable. Depending on the task, the answer may be a temporary manual process or a deliberate pause until the issue is resolved. The appropriate response depends on the consequences of missing, delaying or duplicating the action.
When ownership changes, include the automated process in the handover. Explain its purpose, boundaries, review routine and known limitations. A system that continues operating after everyone has forgotten why it exists is difficult to supervise responsibly.
A worked example of creating review tasks
This is an illustration. A small design business wants completed internal briefs to receive a quality review. At present, the coordinator checks the work list and manually creates a review task for each eligible brief. The step is sometimes delayed when incoming work becomes busy.
The business first clarifies eligibility. A brief must be in the agreed ready-for-review stage, contain the required sections and identify its owner. Moving a brief into that stage is a human decision. The proposed automation creates an internal review task; it does not approve the brief or send anything to a customer.
The team trials the process on a limited set of work. It checks that each eligible brief creates one task, incomplete records are brought to the coordinator’s attention and repeated updates do not create duplicate tasks. The reviewer still examines the substance and records the outcome.
For the hypothetical time comparison, the business handles 20 eligible briefs each month. Creating each task manually takes 3 minutes, giving 60 minutes of monthly administrative work. The new process requires a 15-minute monthly check and, in this example, 10 minutes to resolve exceptions, leaving 35 minutes of net monthly time saved.
Suppose setup and testing take 210 minutes. At a continuing saving of 35 minutes per month, the setup time is recovered after six months. This calculation considers time only. It excludes software costs, future changes and any difference in review timeliness or error rates.
The business should therefore assess more than the arithmetic. If eligible briefs wait less time and missing reviews become easier to detect, the process may have value beyond the administrative saving. If exception work grows or the underlying workflow changes, the original estimate needs to be revised.
Make commitment arrangements fair and useful
Advance commitments work best when people understand what they are agreeing to and can influence the arrangement. A recurring meeting should have a clear output, suitable participants and a realistic allowance of time. Attendance alone is not evidence that the commitment serves its purpose.
Avoid personal penalties or public embarrassment as a substitute for good work design. A person may miss an agreed action because information was late, priorities changed or the workload was unrealistic. Those situations call for investigation and a practical response.
Separate voluntary personal experiments from workplace requirements. A founder might choose a personal focus arrangement, while a shared operating process affects other people and needs broader agreement. Any formal employment obligations require appropriate advice and reference to current rules.
Review whether the commitment is still helping. If cancelling a meeting repeatedly becomes the sensible decision, its timing or purpose may be wrong. The useful response is to redesign the arrangement, rather than preserve it solely because it once seemed like a good commitment.
Notice when convenience points in the wrong direction
Clear also observes that convenient systems can reinforce behaviour people did not deliberately choose. In a business, a tool may make responding to every notification easier than completing planned work. A default may encourage unnecessary distribution of information because the broad recipient list is already selected.
Examine the behaviour that the current setup makes easiest. Compare it with the behaviour the business intends to support. If they differ, a change to notifications, starting settings or workflow may be more useful than repeatedly reminding people to be disciplined.
Introduce friction selectively when a pause improves a decision. A review before a consequential action can be useful, provided the reason is clear and the process does not obstruct routine work unnecessarily. More obstacles are not automatically better controls.
Check the burden on people who use the system differently. A setting that helps one role concentrate may make another role miss time-sensitive work. Agree on role-appropriate arrangements and a clear route for urgent communication before changing shared defaults.
Avoid automating an unclear or unwanted obligation
Automation can make a weak process more consistent. Before implementing it, ask whether the task still needs to happen, whether its frequency is appropriate and whether its output is used. Removing a redundant step may be more valuable than executing it faster.
Watch for processes built around a temporary problem. A recurring report introduced during a short disruption may continue long after the disruption ends. Its regular arrival can give it an appearance of importance even when nobody uses it to decide anything.
Also consider the cost of changing the arrangement later. A heavily connected process can be difficult to understand and revise. Begin with a clear boundary and document the dependencies that matter, rather than connecting every available system at once.
Finally, do not treat the absence of visible complaints as proof of success. Inspect a sample of outcomes and speak with the people receiving them. They may have developed quiet workarounds that hide the cost of an otherwise impressive automation.
Questions to ask
- Which recurring failure are we trying to prevent?
- Would a reminder, clearer default or simpler process be sufficient?
- Are the trigger, required inputs and expected result clearly defined?
- Who handles exceptions and owns the process after setup?
- How will we detect missing, duplicated or unsuitable actions?
- What would cause us to change or retire the arrangement?
Bringing it together
Commitments and automation can support good work by making intended actions easier to sustain. Their value depends on a clear purpose, suitable boundaries and a person who remains responsible for the result.
Begin with one recurring problem. Choose a proportionate mechanism, test it on real work and include exceptions and review from the outset. A dependable arrangement reduces avoidable effort while keeping important decisions visible and open to reconsideration.
Source: James Clear, Atomic Habits (2018), primarily chapter 14. Business methods and examples are practical adaptations. Figures in this article are illustrations, not data. This article is general business information, not legal advice.
