"When I Feel Like It" Never Comes: How to Keep Working Without Relying on Willpower

• Who this is for: Knowledge workers and software engineers who keep thinking “if I could just get motivated, I’d focus,” and then reach the end of the day with that motivation never having shown up
• Prerequisites: None (no background in psychology required)
• Reading time: about 15 min

Overview

When work stalls, the first thing most of us go looking for is an internal switch. Psych yourself up, summon some motivation, exercise discipline. Flip the switch and you can move; leave it off and you can’t. A lot of people build their whole day on that assumption. But the assumption has a scientific weak spot.

The idea that willpower is a fuel that gets used up (ego depletion) became one of the most famous replication failures in 2010s psychology. A large-scale registered replication across 36 labs and 3,531 participants shrank the effect to nearly zero¹. The intuitive model, that you top up your tank of grit and push through, stands on much shakier ground than it feels like it does.

What does have evidence behind it is the opposite move. Behavior is a result of mood, but it is also a cause of it. Behavioral activation, an approach established in depression treatment, flips the usual order: instead of “move once you feel motivated,” it works on “move in a small way and the mood and motivation catch up afterward”²³. And there are several techniques that take the act of getting started out of the hands of willpower and hand it to a system instead. Implementation intentions tie a behavior to a trigger in advance: “when X happens, do Y”⁴. Habits get automated by contextual cues⁵⁶. Choice architecture designs the defaults and the friction in your environment⁷. Precommitment lets you impose a deadline on yourself⁸. Each of these has its own evidence.

The claim of this article is simple. You will get more done if you design for the assumption that motivation cannot be counted on. The point is not to “produce” motivation but to set up, in advance, the conditions under which action happens even when motivation is absent. On top of that, if you can stop carrying externally imposed work as something done under duress and re-read it as a value of your own (what self-determination theory calls internalization⁹), the systems last longer. After checking the collapsed ground under the motivation myth, I’ll walk through the concrete systems for moving by design rather than by willpower, fitting each one into an engineer’s day.

Part 1: The ground under the motivation myth has collapsed

Start by inspecting an assumption most of us hold without question: that tackling something important requires motivation or willpower first.

The science that long propped this up was Roy Baumeister and colleagues’ theory of ego depletion. Willpower was a finite resource that tires like a muscle; perform one act of difficult self-control and the next one fails. Ever since the famous 1998 “cookies and radishes” experiment, the idea spread widely, from self-help to business books. A 2010 meta-analysis (198 tests) reported a moderate effect (d ≈ 0.6), and the theory looked solid¹.

Then in the late 2010s, the theory became one of the emblematic casualties of psychology’s “replication crisis.” Once large-scale preregistered replications ran the same experiment across labs worldwide using a single protocol, the effect vanished. A replication with around 2,000 people found no sign of it, and one mobilizing 36 labs and 3,531 participants produced an effect size of d ≈ 0.06. That is an order of magnitude smaller than the original estimate, effectively zero¹. Some analyses find that once you statistically correct for publication bias, the effect becomes indistinguishable from zero.

Don’t take this the wrong way. This is not “willpower doesn’t exist” or “effort is pointless.” What collapsed is the more specific assumption, the one that actually constrains our behavior the most: the model in which “manage your tank of grit well and you can move / the tank ran dry so you can’t.” If that model is unreliable, then the reasoning behind “no motivation today, so I’ll put it off” and the hope that “I’ll refocus if I just psych myself up again” are both less trustworthy than they feel.

You’ve probably noticed it yourself. The tasks you waited to feel motivated for are the ones that never get started. And the opposite happens too: once you actually touch a task, you often think, “huh, this is moving better than I expected.” That second feeling is exactly what behavioral activation, which we’ll look at next, explains.

Part 2: Action first, mood second

If “move once motivated” can’t be trusted, reverse the order. Move first, and the motivation comes.

It can sound like wishful thinking, but this is the core principle of behavioral activation, an approach established in depression treatment. Formulated by Lewinsohn in the 1970s and Jacobson and colleagues in the 1990s, it focuses on a vicious cycle: not “I’m down so I can’t move” but “I stopped moving, which dragged my mood down further.” So treatment doesn’t wait for mood to recover. It schedules small actions first (take a walk, see someone, clear one task off the list). As activity goes up, mood follows².

The evidence holds up. A meta-analysis pooling multiple randomized trials found behavioral activation significantly outperformed control groups (standardized mean difference of about −0.74), with effect sizes comparable to well-established cognitive behavioral therapy³. What matters more is the mechanism. In some studies that tracked the course of activity within sessions, changes in activity preceded or coincided with changes in mood. People didn’t move because their mood improved; their mood improved because they moved².

flowchart TB
    A["Wait to<br>feel motivated"] --> B["Never<br>gets started"]
    B --> C["The sense that<br>'I can't move'<br>gets stronger"]
    C --> A
    D["Move in a<br>small way<br>(5 min is fine)"] --> E["The barrier drops<br>and you<br>make a little progress"]
    E --> F["Sense of competence<br>and mood<br>tick upward"]
    F --> G["The next action<br>comes more<br>easily"]
    G --> D

One caveat. Behavioral activation is a treatment validated in clinical settings (depression and anxiety). It has not been directly shown to “raise the work productivity of healthy people.” What I’m borrowing here is the principle, the direction that “action can precede mood and drive.” That said, the principle matches everyday experience well, and it’s consistent with the research on implementation intentions and habits below.

In practice, it comes out like this. Drop “I’ll write once I’m motivated” and switch to “just write one line.” If a code review feels like a drag, open only the first file. Stop waiting for your mood to get ready; make the entry point to the action absurdly small and step in first. Motivation is something the foot you’ve already moved brings along with it.

Part 3: Four systems for moving by design, not by willpower

“Move in a small way” is powerful, but summoning even that first step by willpower every single time is draining. From here I’ll go through four techniques that outsource the act of getting started from willpower to a system. The shared idea is to tie behavior to a condition rather than to motivation.

System 1: Implementation intentions, deciding “when, where, what” in advance

The best return on effort comes from implementation intentions. You tie a behavior to a trigger ahead of time in the form “when situation X arises, do Y.” That’s all. You convert a vague goal intention (“I should do this”) into a concrete situation-action rule like “first thing Friday morning, write next week’s design memo.”

It looks modest, but the numbers are large. In a meta-analysis by Gollwitzer and Sheeran pooling 94 experiments and about 8,000 people, implementation intentions showed a medium-to-large effect on goal attainment (d ≈ 0.65)⁴. Once a cue and a behavior are linked in advance, the moment that situation arrives the action fires off without leaving room to wonder “what should I do.” You’ve moved the cost of deciding off the floor of execution and back into the planning stage.

For engineers, that looks like:

• When standup ends, the first thing you do is continue yesterday’s hardest task (don’t dissolve the start of the morning in light busywork)
• When a Slack notification catches your eye, you don’t open it until the current Pomodoro ends (handling interruptions in advance)
• When you open a PR, you add one test on the spot (chaining an easily-deferred task to the action)

System 2: Habits, handing the behavior over to contextual cues

Repeat an implementation intention in the same context and it eventually becomes a habit. What habit research tells us is that a habit is not “strength of will” but “an association between a context cue and a behavior.” According to the classic review by Wood and Neal, when a behavior is repeated in a fixed situation, control shifts from conscious intention to environmental cues, and the cue triggers the behavior automatically⁶. What gets you to make coffee is less “I want caffeine” than the combination of cues: the break room, your colleagues, the machine.

So how long does habit formation take? In Lally and colleagues’ study, it took an average of 66 days for a new behavior to become automatic. The spread was wide, though: a simple behavior (drinking water) took 18 days, while a more effortful one was projected to reach 254 days⁵. The most useful finding for practice is the other one. Missing a single day doesn’t seriously derail the progress toward automaticity⁵. “Skip once and it’s ruined” is a myth; if you skip occasionally, just restart matter-of-factly. A long, loose streak builds habits better than a perfect, unbroken one.

The implementation trick is to attach the new habit to an existing cue. “When I open the editor, the first thing I do is clear one TODO comment” uses a context that already happens every day (launching the editor) as the trigger. Riding on an existing routine sticks better than building a cue from scratch.

System 3: Friction and defaults, designing the environment instead

This is the most powerful way to change behavior without using willpower at all. Change the defaults and the friction in your environment. People follow defaults to a surprising degree. In Johnson and Goldstein’s well-known study, organ-donor registration rates in countries where donation is the default (opt-out) were roughly double those in countries where you choose donation yourself (opt-in). Attitudes were similar; the difference in defaults alone moved behavior a lot⁷.

You can apply this directly to your own work environment. Lower the friction on the behaviors you want, and raise it on the ones you want to stop.

Behavior you want	Design that lowers friction
Start focused work in the morning	Leave the editor and branch open the night before
Exercise or take a walk	Set workout clothes next to your work clothes / take meetings while walking
Read or study	Leave the book you want open on your desk

Behavior you want to stop	Design that raises friction
Mindless social media	Log out of the apps / remove them from the home screen
Interruptions from notifications	Physically turn off notifications during focus time / put the phone in another room
Impulsive task-switching	Close unrelated tabs and channels in advance

The key is not to lean on in-the-moment self-control. Friction that makes logging in take 30 seconds beats “the willpower not to look” at staying off social media. Cut the number of decisions themselves, and make the desirable behavior the default that happens when you do nothing.

System 4: Precommitment, binding your future self

Last is the technique where your present self binds the future self who will procrastinate. In Ariely and Wertenbroch’s experiment, students were given the chance to set their own deadlines for three papers. Even though missing them carried a penalty, many set deadlines earlier than the end of term on their own. Self-imposed deadlines reduced procrastination and improved grades. They just weren’t as effective as deadlines optimally imposed from outside⁸.

There are two implications. First, people are aware of their own procrastination habit and want a mechanism to bind it. Second, self-imposed deadlines work, but deadlines that involve other people or external structures are stronger. So don’t lean only on a “promise to yourself”; add some externality and it works.

• Tell a colleague “I’ll have the review out Friday” (social commitment)
• Schedule pair or mob programming in advance (binding via time and other people)
• Set a cost on failure, like “donate if I don’t make it”

Part 4: Killing the sense of being forced

The four systems so far were about designing the start of action. But to keep a system running for a long time, you need one more thing: whether you can take on the work as your own.

Self-determination theory sorts extrinsic motivation into qualitatively different stages⁹. There’s “external regulation,” grudgingly doing something for reward or punishment; “introjected regulation,” doing it out of guilt or pressure; “identified regulation,” doing it because you understand the activity’s significance as your own value; and “integrated regulation,” which is fused with how you see yourself. The quality and durability of motivation change with how far you’ve internalized an externally given goal into your own values.

This is where it connects to the “don’t rely on motivation” thread. External and introjected regulation are easily swayed by the waves of mood and pressure. Motivation internalized all the way to identification, on the other hand, depends little on whether you feel like it. The moment you can re-read “the boring migration work the boss told me to do” in your own words as “this migration cuts rework for the whole team next quarter,” the action gets decoupled from the weather of your mood.

This “editing of meaning” is powerful precisely because you can generate motivation by changing how you see the work, without changing the work itself. In organizational psychology it’s one facet of job crafting (cognitive crafting). I covered how editing the meaning of work moves well-being and engagement, with the effect sizes and the pitfalls (quietly cutting your own chores shifts the burden onto colleagues, for instance), in a separate article. The one point worth holding in this context: once you’ve automated the start with a system and can connect the work to your own values, the system turns from a drain into something that runs on its own.

Part 5: Dropping it into an engineer’s day

Let me turn the abstract into something concrete you can use tomorrow. A “day that doesn’t rely on motivation” can be designed something like this.

The night before (lower the friction)

Decide the one task you’ll tackle first thing in the morning, and close the laptop with the editor and branch already open. Get the “what do I do in the morning” decision out of the way the night before.

Morning (start with an implementation intention)

“When standup ends, before busywork or Slack, touch yesterday’s hardest task for just 5 minutes.” Five minutes is enough. As the behavioral activation principle says, make the entry point absurdly small. Once you’re moving, you usually keep going.

During the day (defend with habits and friction)

Physically turn off notifications during focus blocks. Close social media and unrelated tabs (raise the friction on behaviors you want to stop). Make a chain like “when I open a PR, I add one test on the spot” a habit by attaching it to an existing cue (creating the PR).

Deadlines (precommitment)

Don’t finish important tasks on a “deadline inside your own head.” Pull in other people by booking a review or a pairing session.

Now and then (check the meaning)

For the tasks that feel most boring or imposed, write out in one sentence “whose what does this make easier.” Once the meaning is internalized, you become resistant to the waves of mood.

Notice anything? Not one of these steps involves “psyching yourself up.” There’s no attempt to produce motivation. You’re just arranging, in advance, the conditions under which action happens even when motivation is absent.

Caveats: not a cure-all

In fairness, here are the limits.

First, individual variation is large. The time needed to form a habit ranges from 18 to 254 days⁵, and which systems work differs from person to person. What I’ve listed is a menu, not a prescription. Try things and keep what works for you.

Second, behavioral activation was originally validated in clinical settings (depression and anxiety); it has not directly proven productivity gains in healthy people. This article borrows it at the level of principle. Don’t overstate it.

Third, systems are indifferent to the goal. There’s a real danger of running efficiently in the wrong direction. “Automating the start of action” has to come paired with the work of correcting what you’re moving toward (meaning, prioritization), or it risks ending in nothing but busyness.

Fourth, the model that treats willpower as “fuel” lost its footing in replication¹, but the phenomena of self-control and effort didn’t disappear. The point of this article isn’t “don’t try.” It’s this: don’t spend your total budget of effort on mood-dependent starts every single day; invest it in a one-time design.

Summary

“I’ll get serious once I’m motivated.” The weakness of that strategy comes down to one thing: the arrival of motivation isn’t guaranteed. The model that treats willpower as refillable fuel failed to replicate in large-scale studies¹. Building your day on something unreliable is a fragile design.

What you can use instead is reversing the order and outsourcing to a system. Because behavior can precede mood, don’t wait for motivation; make the entry point small and step in first²³. Don’t leave the start to willpower; tie behavior to a condition rather than motivation, using implementation intentions⁴, habits⁵⁶, friction and default design⁷, and precommitment⁸. On top of that, if you can internalize the work as your own value⁹, the system runs on its own instead of draining you.

Motivation isn’t something you produce; it’s something whose conditions you create. Those conditions, once designed on a day when you happen to feel like it, will move you on the days you don’t. My own view is that this is the most useful property of the whole thing.

Related articles

You may also be interested in these related posts:

• Good Work Isn’t Chosen—It’s Crafted: The Evidence Behind the Seven Factors of Job Satisfaction - A deeper look at the effect sizes and pitfalls of “editing the meaning” = job crafting
• Psych Self-Care, Ranked by Effect Size: A Catalog of 8 Evidence-Based Techniques for a Tired Mind - Compares self-care techniques, including if-then implementation intentions, by effect size
• The Comfort Crisis: Four Discomforts Knowledge Workers Must Reclaim - The view of not just “lowering” friction but deliberately adding it

References

References are listed in order of the citation numbers used in the text.

1. A Multilab Registered Replication Report on the Ego-Depletion Effect - Vohs, K. D., et al., Psychological Science, 32(10), 1566-1581 (2021). 36 labs, 3,531 participants, d ≈ 0.06. Contrasted with Hagger et al. (2010) meta-analysis, d ≈ 0.6. Reliability: High. ↩︎ ↩︎² ↩︎³ ↩︎⁴ ↩︎⁵

2. A Narrative Review of Empirical Literature of Behavioral Activation Treatment for Depression - Frontiers in Psychiatry (2022). The “action precedes mood” principle, and observations that changes in activity precede changes in mood. Reliability: Medium to High. ↩︎ ↩︎² ↩︎³ ↩︎⁴

3. Behavioural Activation for Depression: An Update of Meta-Analysis of Effectiveness and Sub Group Analysis - Ekers, D., et al., PLoS ONE (2014). SMD ≈ -0.74 versus control. Reliability: High. ↩︎ ↩︎² ↩︎³

4. Implementation Intentions and Goal Achievement: A Meta-Analysis of Effects and Processes - Gollwitzer, P. M., & Sheeran, P., Advances in Experimental Social Psychology, 38, 69-119 (2006). 94 experiments, about 8,000 people, d ≈ 0.65. Reliability: High. ↩︎ ↩︎² ↩︎³

5. How are habits formed: Modelling habit formation in the real world - Lally, P., et al., European Journal of Social Psychology, 40, 998-1009 (2010). Average of 66 days to automaticity (range 18-254 days); a single missed day has no serious effect. Reliability: High. ↩︎ ↩︎² ↩︎³ ↩︎⁴ ↩︎⁵

6. A New Look at Habits and the Habit–Goal Interface - Wood, W., & Neal, D. T., Psychological Review, 114, 843-863 (2007). Habit = an association between context cue and behavior. Reliability: High. ↩︎ ↩︎² ↩︎³

7. Do Defaults Save Lives? - Johnson, E. J., & Goldstein, D. G., Science, 302, 1338-1339 (2003). Opt-out organ-donor registration rates roughly double those of opt-in. Reliability: High. ↩︎ ↩︎² ↩︎³

8. Procrastination, Deadlines, and Performance: Self-Control by Precommitment - Ariely, D., & Wertenbroch, K., Psychological Science, 13, 219-224 (2002). Self-imposed deadlines reduce procrastination but are not as effective as deadlines optimally imposed from outside. Reliability: High. ↩︎ ↩︎² ↩︎³

9. Self-Determination Theory and the Facilitation of Intrinsic Motivation, Social Development, and Well-Being - Ryan, R. M., & Deci, E. L., American Psychologist, 55, 68-78 (2000). The continuum of extrinsic motivation and internalization (identified and integrated regulation). Reliability: High. ↩︎ ↩︎² ↩︎³