(Behavioural Science) #14 Commitment Devices

Principle #14 · Pre-commitment / Choice architecture

Commitment devices

A commitment device is a voluntary choice made in advance — before temptation arrives — that restricts your future options or raises the cost of backing down. By binding the future self when motivation is high, it offsets the predictable drop in willpower or resolve at the moment of action.

2×

savings increase in Ashraf et al. Philippines study

29%

higher gym attendance with pre-commitment in Acland & Levy

3x

more smokers quit using financial commitment contracts vs. control

~35%

of people offered a commitment take it up when incentive is real

1. What it is and the science behind it

The core problem commitment devices solve is present bias — the tendency to overweight immediate costs and rewards relative to future ones. When you set your alarm the night before, you are in a "cold state," thinking clearly about tomorrow's goals. When the alarm goes off, you are in a "hot state," and a different calculus takes over. Commitment devices are a mechanism for letting the cold-state self constrain the hot-state self.

The theoretical foundation comes from Thaler and Shefrin's planner-doer model (1981) and Strotz's earlier work on dynamic inconsistency (1955). The empirical revolution came when economists started running field experiments to test whether people would actually use these devices — and how much they helped.

Why it works — three mechanisms

Raises the cost of failureFinancial stakes, public pledges, or social accountability make backing out more painful than following through. The future-self calculation shifts.

Removes the decisionPre-committing eliminates the need to re-decide at the moment of temptation. Willpower is never called on because there is no live choice.

Signals identityThe act of committing sends a signal to yourself (and others) about who you are. Backing out feels like a violation of identity, not just a missed goal.

Key studies

SEED savings account — Philippines

Ashraf, Karlan & Yin, 2006

Low-income rural bank customers were offered a "commitment savings account" with a self-chosen lock-in: either a date lock (cannot withdraw before X date) or a goal lock (cannot withdraw until balance reaches $Y). Those who took up the account saw savings balances grow by 81% more over 12 months than the control group — roughly doubling savings rates. Critically, the account paid no interest premium; the commitment itself was the value.

+81% savings growth vs. control

Stickk — anti-charity commitment contracts

Ariely & Wertenbroch, 2002; platform-scale replications 2010–2018

Participants stake real money against a goal (weight loss, exercise, quitting smoking). If they fail, the money goes to a cause they dislike — an "anti-charity" — intensifying the loss aversion effect. In smoking cessation studies using this design, quit rates were approximately 3x higher than control groups offered only information and social support. The financial stake was more predictive of success than the size of the stake.

3× higher quit rate vs. control

Gym attendance and pre-commitment

Acland & Levy, 2015

New gym members who were offered a subsidized commitment contract (pay now to lock in attendance targets, get a refund for hitting them) attended significantly more in the following month than those given standard memberships. However, the effect attenuated over time — suggesting that commitment devices work best for bridging the intention-action gap in early habit formation, but need to be restructured as habits consolidate.

+29% attendance in first month

Save More Tomorrow (SMarT) — retirement savings

Thaler & Benartzi, 2004

Employees who could not bear to cut take-home pay today were offered an alternative: commit now to directing a portion of future raises into retirement savings. Because the contribution increases were tied to pay raises, people never experienced a nominal decline. Average savings rates in companies that adopted SMarT rose from 3.5% to 13.6% over 40 months — a near-quadrupling. This became the most empirically influential commitment device design in policy history.

3.5% → 13.6% savings rate

2. Real application examples

Business

SMarT-style auto-escalation

Pension providers offer employees automatic contribution increases tied to future pay rises. Once enrolled, inertia keeps the commitment alive. Used by major 401(k) providers globally post-Pension Protection Act 2006.

Business

Pre-ordering and pre-payment

SaaS companies offer annual upfront plans at a discount. Customers commit to 12 months when motivation is high at sign-up, bypassing the monthly cancellation decision. Reduces churn by removing the recurring choice point.

Business

Website blockers with scheduling

Tools like Freedom or Cold Turkey let users pre-schedule distraction blocks — locked in the night before, when willpower is intact. The future self cannot override them without significant friction (e.g. 24-hour waiting period).

Public policy

Commitment savings for low income

Following the SEED study, the Philippines and Kenya integrated locked savings products into national financial inclusion programs. Similar products in Kenya (M-Pesa goal accounts) show large uptake and measurable savings increases among households previously excluded from formal banking.

Public policy

Advance market commitments

Governments commit in advance to purchasing a vaccine or technology at a guaranteed price if it meets a specification. This removes R&D risk for manufacturers, enabling investment that would not occur without the commitment signal. Used for pneumococcal vaccines and COVID-19 procurement.

Public policy

Drink-drive pledges at checkpoints

Police and health campaigns ask drivers at sober checkpoints to sign a pledge card committing to not drink and drive. Studies in Thailand and the UK show modest but significant reductions in repeat drink-driving, especially when the pledge includes a named witness.

Personal habit

Paying a deposit for a class

Pre-paying for a yoga class, personal training session, or online course creates a financial commitment. The sunk cost effect combines with commitment to substantially increase actual attendance vs. free or pay-later options.

Personal habit

Public declarations

Announcing a goal to friends, family, or social media adds a social accountability layer. Research shows public commitment outperforms private commitment — the fear of social judgment raises the cost of failure. Charity fundraising pages (e.g. running a marathon) combine public commitment with financial stakes.

Personal habit

Ulysses contracts in healthcare

Patients in stable states write advance directives specifying how they want to be treated if they become incapacitated or enter a crisis. Used in psychiatric care for patients with bipolar disorder — the well-self pre-commits to treatment acceptance before a manic episode can override consent.

3. Design guidance — when and how to use it

When this principle works well

Use when

There is a clear gap between what people intend (cold state) and what they do (hot state) — the classic present-bias problem.

Use when

The target behavior is measurable and verifiable — commitment only works if you can objectively check whether the person followed through.

Use when

The person has genuine pre-existing motivation but weak self-control. Commitment devices amplify existing intent — they cannot create intent from scratch.

Use when

A "fresh start" moment is available — beginning of the year, new job, post-holiday. Motivation is naturally higher, making commitment uptake more likely.

Avoid when

The target behavior is ambiguous or involves significant judgment calls — rigid commitment can prevent adaptive response to new information.

Avoid when

The stakes are asymmetric and irreversible. Commitment to a financial penalty for someone already financially fragile can cause real harm and trigger shame-driven disengagement.

Four types of commitment device — choose by context

Financial stakes

Put real money on the line — forfeited to charity, an anti-charity, or another party if the commitment is broken.

Best for: health behaviors, savings goals, productivity. Strongest effect size but lowest uptake — only ~35% of people offered a monetary commitment take it.

Social accountability

Public pledge, named witness, or partner agreement. Social consequences (embarrassment, reputation) replace financial ones.

Best for: exercise, diet, learning goals. Higher uptake than financial commitments; works especially well for collectivist cultures.

Access restriction

Physically removing the option — blocking a website, locking away a credit card, pre-portioning food.

Best for: digital overuse, impulse spending, dietary behavior. Zero enforcement cost; most durable effect because it eliminates the choice entirely.

Pre-commitment scheduling

Booking the appointment, buying the ticket, putting it in the calendar before motivation wanes.

Best for: healthcare, exercise, professional development. Low friction, high scalability — easiest to build into product or service flows.

Step-by-step design process

1. Identify the cold-state moment — find the point at which motivation and clarity are highest. This is when to offer the commitment option. For savings: enrollment. For health: post-diagnosis. For learning: course sign-up.
2. Make the commitment specific and measurable — vague pledges ("I'll be healthier") have weak effect. Effective commitments specify what, how much, and by when. "I will exercise 3 times per week for 8 weeks" outperforms "I'll exercise more."
3. Choose the right type of device for the context and audience — financial stakes offer the strongest effect but lowest uptake; scheduling and access restriction offer better scale; social accountability works when social norms are strong.
4. Add a verification mechanism — the commitment needs a third party, a technology, or a social witness to hold the person accountable. Self-reporting alone reduces effect size significantly.
5. Design an easy off-ramp with deliberate friction — a completely inescapable commitment can backfire, creating resentment or learned helplessness. The goal is meaningful friction, not a trap. Allow exit with delay and cost, not instant reversal.
6. Pair with identity framing — the commitment will be stronger if it is framed as an expression of who the person is, not just what they want to do. "People who care about their family's future start saving today" outperforms "Save now for better returns."

Before and after — message design

Retirement savings enrollment

Weak

"Consider increasing your pension contributions when you can afford to."

Strong (SMarT)

"You don't have to save more now. Just commit to putting half of your next raise into your pension. You'll never see your take-home pay go down."

Exercise goal

Weak

"Set yourself a fitness goal for the new year."

Strong

"Book your first 3 sessions now — they're refundable if you cancel 24h before. Most people who book ahead show up. Which days work best for you this week?"

Healthy eating

Weak

"Try to make healthier choices at lunch."

Strong (pre-order)

"Order tomorrow's lunch before 10am today. People who pre-order their meals choose 31% fewer calories on average — and they feel just as satisfied."

Critical ethical boundary

Commitment devices must be voluntary and genuinely chosen by the person in their own interest. Designing a commitment that benefits the company at the person's expense — e.g. automatic annual renewal with high cancellation friction — is not a nudge, it is a trap. The ethical test: would the person, fully informed and in a clear-headed moment, choose this commitment for themselves? If the answer is no, redesign. The most powerful long-term commitment designs are those the person is glad to be held to — not ones they feel tricked by.