Abstract

People do not always do what they intend to do (Sheeran & Webb, 2016). This article reviews four bodies of evidence on why starting a task is hard and what helps it begin. First, procrastination is most consistently linked to aversive tasks, delayed rewards and impulsiveness, and it trades short-term relief for later costs (Steel, 2007; Tice & Baumeister, 1997; Sirois & Pychyl, 2013). Second, intentions for the future are often lost at the moment to act, and external reminders set in advance improve their fulfilment (Einstein & McDaniel, 1990; Gilbert, 2015; Rogers & Milkman, 2016). Third, field experiments find that reminders and prompts to plan a specific date and time raise follow-through in vaccination, saving and exercise, with modest effects (Milkman et al., 2011; Karlan et al., 2016; Calzolari & Nardotto, 2017; Milkman et al., 2021). Fourth, defaults show that when an action happens unless people intervene, inertia works for it, although the size of such effects is contested (Madrian & Shea, 2001; Jachimowicz et al., 2019; Maier et al., 2022). On this evidence, Mars lets a reminder carry a prompt that it starts running in a new conversation at the set time, saves the meetings you mention in chat as events, and sends event alerts to your bound WeChat. We also state the limits of this evidence: it supports the design direction and is not an experimental test of Mars itself.

Keywords: procrastination · task aversiveness · prospective memory · intention offloading · reminders · default effects · choice architecture · AI reminder assistant

1. Introduction

Procrastination has costs. In two longitudinal studies of students, procrastinators reported less stress and illness than others early in the semester but more late in the term, were sicker overall, and received lower grades on all assignments (Tice & Baumeister, 1997). A meta-analytic review described procrastination as a prevalent form of self-regulatory failure whose prevalence appears to be growing (Steel, 2007). More broadly, research on intention–behaviour relations converges on the conclusion that people do not always do the things they intend to do (Sheeran & Webb, 2016).

This article asks why the start of a task is so often where an intention fails, which conditions help a planned action actually begin, and how those answers translate into the design of an assistant that handles reminders.

Method: this is a narrative literature review. We favoured peer-reviewed work accessible through a DOI, and preferred meta-analyses, longitudinal studies and randomized field experiments. Where a popular claim is contested we cite the critique, and where a correction was published we use the corrected figures. Every factual statement carries its source, and the full reference list with original links follows the article.

2. Why starting is hard

Steel’s meta-analysis of 691 correlations found that neuroticism, rebelliousness and sensation seeking had only a weak connection with procrastination, while task aversiveness, task delay, self-efficacy and impulsiveness, together with conscientiousness and facets such as self-control, distractibility and organization, were strong and consistent predictors (Steel, 2007). These results fit temporal motivation theory, which combines expectancy theory with hyperbolic discounting, the tendency for rewards that arrive later to count for less at the moment of choice (Steel, 2007).

Sirois and Pychyl argue that procrastination often occurs when a task is seen as aversive, and that it has a great deal to do with short-term mood repair: putting the task off relieves the present self, while the consequences are borne by the future self (Sirois & Pychyl, 2013). Activities with up-front costs and delayed benefits, such as getting an education or maintaining a healthy lifestyle, have this structure (Calzolari & Nardotto, 2017).

The cost of a task is therefore concentrated at its start, when the effort is immediate and the benefit furthest away. Based on this evidence, Mars lets you add a prompt when you create or edit a reminder: at the set time, Mars opens a new conversation and runs it, and you can add a Daily review skill right from the reminder editor. An alarm only tells you to begin; with a prompt attached, opening the conversation and writing the request are already done when the time comes (Steel, 2007; Sirois & Pychyl, 2013).

3. Intentions slip at the moment to act

Remembering to do something at a future moment is a prospective memory task. In a laboratory paradigm comparing younger and older adults, Einstein and McDaniel found that external aids and unfamiliar target events benefit prospective memory performance (Einstein & McDaniel, 1990). Later experiments supported the view that such retrieval can rely either on controlled monitoring of the environment for a target event or on spontaneous retrieval when the event appears, and that people rely on each to different degrees depending on task demands and individual differences (Einstein et al., 2005).

People also move intentions out of their heads. In four online experiments with 1,196 participants, Gilbert found that setting external reminders improved performance, and that participants set reminders adaptively, according to memory load and the likelihood of distraction (Gilbert, 2015).

Not every reminder works equally well. Rogers and Milkman found that cue-based reminders are more potent when the cue is distinctive relative to other stimuli and captures attention when there is an opportunity to act; such reminders can be more effective than written or electronic reminder messages, and people undervalue and underuse them (Rogers & Milkman, 2016).

Mars applies this to plans that people mention in passing. Tell Mars about an upcoming meeting in chat and it saves the event and links the participants, so the same record shows on each person’s Events tab, on the web and iOS, with nothing to enter twice; plans and dates mentioned in a chat become events on your calendar, with reminders before the moments that matter. The intention leaves memory without a separate step of writing it down (Gilbert, 2015; Einstein & McDaniel, 1990).

4. What reminders do in the field

Field experiments show both the reach and the size of reminder effects. At a large firm, all eligible employees received mailings listing the times and locations of free on-site vaccination clinics; those also prompted to write down the date and time they planned to be vaccinated had a vaccination rate 4.2 percentage points higher than the control rate of 33.1%, while a prompt for the date alone gave a 1.5-point difference that was not statistically significant (Milkman et al., 2011).

In field experiments with three banks, reminder messages increased commitment attainment among clients who had recently opened commitment savings accounts; messages that mentioned both the savings goal and the financial incentive were particularly effective, while gain-versus-loss framing made no significant difference (Karlan et al., 2016). At a gym, simple weekly reminders substantially increased attendance over an extended period; the response came within hours and recurred with each reminder, consistent with limited attention, and some habit persisted after the reminders ended, though with a smaller effect (Calzolari & Nardotto, 2017).

Health care offers the largest samples. A meta-analysis of text-message (SMS) appointment reminders found a summary effect of 1.48 (95% CI 1.23–1.72) on attendance across eight randomized trials, with no significant subgroup differences by message timing of 24, 48 or 72 or more hours before the appointment (Guy et al., 2012). A megastudy of 47,306 patients testing 19 text messages found that texts sent before a primary care visit raised flu vaccination by an average of 5%; messages did better when framed as reminders of a vaccine already reserved for the patient and when they matched what patients expected to hear from their provider (Milkman et al., 2021).

Two design lessons follow: a reminder should state concretely what to do and when (Milkman et al., 2011; Karlan et al., 2016), and short messages are a well-tested way to deliver it (Guy et al., 2012; Milkman et al., 2021). Following this, a Mars reminder ties a time to a prompt that says what should happen; turn on WeChat reminders under Settings > Notifications on the web or iOS and every event alert also goes to your bound WeChat, with no extra app to watch.

5. Making the start the default

Behavioural economics adds another lever: change what happens when people do nothing. Under automatic enrollment, 401(k) participation was significantly higher, and a substantial fraction of participants kept both the default contribution rate and the default fund allocation, which the authors attribute to inertia and to employees perceiving the default as investment advice (Madrian & Shea, 2001).

A meta-analysis of 58 default studies (pooled n = 73,675) found a considerable effect, d = 0.68 (95% CI 0.53–0.83), but with substantial variation: several studies found no significant effect and two found negative effects (Jachimowicz et al., 2019). Across more than 200 studies of choice architecture, a meta-analysis reported, after its published correction, an overall effect of d = 0.43 (95% CI 0.38–0.48); interventions that change the structure of the decision outperformed those that describe the alternatives or reinforce behavioural intentions, and the authors found a moderate publication bias toward positive results (Mertens et al., 2022).

That overall estimate is contested. Re-analysing the same data with a robust Bayesian correction for publication bias, Maier and colleagues found no evidence for an overall effect of nudging, evidence against an effect for information and assistance interventions, undecided evidence for structural ones, and heterogeneity implying that some nudges might still be effective (Maier et al., 2022).

A default changes what happens without asking for effort; a reminder that runs by itself applies a similar idea to starting a task. Based on this, Mars makes the start of a routine task what happens at the set time: write the prompt once, and you never have to kick it off by hand again (Madrian & Shea, 2001; Jachimowicz et al., 2019). Because the size of such effects is contested, we treat this as a design direction, not a promised effect (Maier et al., 2022).

6. From evidence to design

These findings point to one conclusion: an assistant that helps people get started should take the first step at the agreed time, keep intentions outside memory, deliver concrete reminders where people will see them, and make the start the default rather than one more decision. Mars meets each requirement as follows:

1. Starting is aversive and its reward is distant → reminders that start the workAdd a prompt when you create or edit a reminder and, at the set time, Mars opens a new conversation and runs it; you can also add a Daily review skill right from the reminder editor. The first step of a routine task is already taken when the moment comes (Steel, 2007; Sirois & Pychyl, 2013).
2. Intentions are lost without a cue → plans from chat become eventsTell Mars about an upcoming meeting in chat and it saves the event and links the participants, so the same record shows on each person’s Events tab, with reminders before the moments that matter. The intention leaves memory without an extra step (Einstein & McDaniel, 1990; Gilbert, 2015).
3. Concrete reminders work as messages → event alerts in WeChatTurn on WeChat reminders under Settings > Notifications on the web or iOS and every event alert also goes to your bound WeChat; scan the QR code under Settings > Bind WeChat to talk to Mars right in WeChat (Guy et al., 2012; Milkman et al., 2021).
4. Inertia favours what happens by default → the start happens at the set timeWrite the prompt once and Mars starts it at the set time, so a routine no longer depends on remembering to begin. The prompt and the time are yours to write and set (Madrian & Shea, 2001; Jachimowicz et al., 2019).

7. Limitations

First, the studies cited here test general regularities of human behaviour, not Mars itself. They support our design direction but do not prove that using Mars produces the same effects; the product’s effect needs its own evaluation.

Second, several studies draw on specific populations, such as students (Tice & Baumeister, 1997), employees of one large firm (Milkman et al., 2011), clients of three banks (Karlan et al., 2016), members of one gym (Calzolari & Nardotto, 2017) and patients of health services (Guy et al., 2012; Milkman et al., 2021), so generalizing to other cultures and tasks calls for care. The effects are modest, for example 4.2 percentage points (Milkman et al., 2011) or an average of 5% (Milkman et al., 2021), and the procrastination evidence rests largely on correlations (Steel, 2007).

Third, the size of nudge effects is contested. After correction for publication bias, no evidence for an average effect remains (Maier et al., 2022); default effects vary widely between studies (Jachimowicz et al., 2019); and the meta-analysis being re-analysed itself reported publication bias and published a correction (Mertens et al., 2022). We cite this work for direction, not for an expected effect size.

Fourth, more reminders are not always better. In the savings experiments, there was no evidence that additional late reminders had an additive effect (Karlan et al., 2016). Among 112 primary care clinicians, the likelihood of accepting a clinical reminder fell by 30% for each additional reminder per encounter and by 10% for each five-point rise in the share of repeated reminders, although newly deployed reminders did not lose response over time (Ancker et al., 2017). Text-message programs with individualized or decreasing message frequency were more successful than those with a fixed frequency (Head et al., 2013), and distinctive cues can outperform electronic reminder messages (Rogers & Milkman, 2016). In Mars, a reminder that starts a conversation runs the prompt you wrote, event alerts follow the events saved from what you mention, and WeChat reminders are a switch you turn on under Settings > Notifications.

8. Conclusion

Starting is where intentions most often stall: the effort is immediate and the reward later, the intention slips from memory at the moment to act, and doing nothing is the path of least resistance. The research points to matching remedies: cues set in advance that arrive at the moment to act, reminders that state concretely what to do, and arrangements in which the first step happens by default. Mars reminders that start a conversation from your prompt, events saved from chat and event alerts in WeChat are designed on this evidence, within the limits stated above.

References

Listed in APA 7th edition style, each with its original DOI link.

1. Ancker, J. S., Edwards, A. M., Nosal, S. C., Hauser, D., Mauer, E. A., & Kaushal, R. (2017). Effects of workload, work complexity, and repeated alerts on alert fatigue in a clinical decision support system. BMC Medical Informatics and Decision Making, 17(1), 36. https://doi.org/10.1186/s12911-017-0430-8
2. Calzolari, G., & Nardotto, M. (2017). Effective reminders. Management Science, 63(9), 2915–2932. https://doi.org/10.1287/mnsc.2016.2499
3. Einstein, G. O., & McDaniel, M. A. (1990). Normal aging and prospective memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 16(4), 717–726. https://doi.org/10.1037/0278-7393.16.4.717
4. Einstein, G. O., McDaniel, M. A., Thomas, R. C., Mayfield, S., Shank, H., Morrisette, N., & Breneiser, J. E. (2005). Multiple processes in prospective memory retrieval: Factors determining monitoring versus spontaneous retrieval. Journal of Experimental Psychology: General, 134(3), 327–342. https://doi.org/10.1037/0096-3445.134.3.327
5. Gilbert, S. J. (2015). Strategic offloading of delayed intentions into the external environment. Quarterly Journal of Experimental Psychology, 68(5), 971–992. https://doi.org/10.1080/17470218.2014.972963
6. Guy, R., Hocking, J. S., Wand, H. C., Stott, S., Ali, H., & Kaldor, J. M. (2012). How effective are short message service reminders at increasing clinic attendance? A meta-analysis and systematic review. Health Services Research, 47(2), 614–632. https://doi.org/10.1111/j.1475-6773.2011.01342.x
7. Head, K. J., Noar, S. M., Iannarino, N. T., & Grant Harrington, N. (2013). Efficacy of text messaging-based interventions for health promotion: A meta-analysis. Social Science & Medicine, 97, 41–48. https://doi.org/10.1016/j.socscimed.2013.08.003
8. Jachimowicz, J. M., Duncan, S., Weber, E. U., & Johnson, E. J. (2019). When and why defaults influence decisions: A meta-analysis of default effects. Behavioural Public Policy, 3(2), 159–186. https://doi.org/10.1017/bpp.2018.43
9. Karlan, D., McConnell, M., Mullainathan, S., & Zinman, J. (2016). Getting to the top of mind: How reminders increase saving. Management Science, 62(12), 3393–3411. https://doi.org/10.1287/mnsc.2015.2296
10. Madrian, B. C., & Shea, D. F. (2001). The power of suggestion: Inertia in 401(k) participation and savings behavior. The Quarterly Journal of Economics, 116(4), 1149–1187. https://doi.org/10.1162/003355301753265543
11. Maier, M., Bartoš, F., Stanley, T. D., Shanks, D. R., Harris, A. J. L., & Wagenmakers, E.-J. (2022). No evidence for nudging after adjusting for publication bias. Proceedings of the National Academy of Sciences, 119(31), e2200300119. https://doi.org/10.1073/pnas.2200300119
12. Mertens, S., Herberz, M., Hahnel, U. J. J., & Brosch, T. (2022). The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains. Proceedings of the National Academy of Sciences, 119(1), e2107346118. https://doi.org/10.1073/pnas.2107346118
13. Milkman, K. L., Beshears, J., Choi, J. J., Laibson, D., & Madrian, B. C. (2011). Using implementation intentions prompts to enhance influenza vaccination rates. Proceedings of the National Academy of Sciences, 108(26), 10415–10420. https://doi.org/10.1073/pnas.1103170108
14. Milkman, K. L., Patel, M. S., Gandhi, L., Graci, H. N., Gromet, D. M., Ho, H., Kay, J. S., Lee, T. W., Akinola, M., Beshears, J., Bogard, J. E., Buttenheim, A., Chabris, C. F., Chapman, G. B., Choi, J. J., Dai, H., Fox, C. R., Goren, A., Hilchey, M. D., … Duckworth, A. L. (2021). A megastudy of text-based nudges encouraging patients to get vaccinated at an upcoming doctor’s appointment. Proceedings of the National Academy of Sciences, 118(20), e2101165118. https://doi.org/10.1073/pnas.2101165118
15. Rogers, T., & Milkman, K. L. (2016). Reminders through association. Psychological Science, 27(7), 973–986. https://doi.org/10.1177/0956797616643071
16. Sheeran, P., & Webb, T. L. (2016). The intention–behavior gap. Social and Personality Psychology Compass, 10(9), 503–518. https://doi.org/10.1111/spc3.12265
17. Sirois, F. M., & Pychyl, T. A. (2013). Procrastination and the priority of short-term mood regulation: Consequences for future self. Social and Personality Psychology Compass, 7(2), 115–127. https://doi.org/10.1111/spc3.12011
18. Steel, P. (2007). The nature of procrastination: A meta-analytic and theoretical review of quintessential self-regulatory failure. Psychological Bulletin, 133(1), 65–94. https://doi.org/10.1037/0033-2909.133.1.65
19. Tice, D. M., & Baumeister, R. F. (1997). Longitudinal study of procrastination, performance, stress, and health: The costs and benefits of dawdling. Psychological Science, 8(6), 454–458. https://doi.org/10.1111/j.1467-9280.1997.tb00460.x

Cite this article

Mars Research. (2026). Why starting is the hardest part: procrastination, reminders and start-up cost. MarsDataAI. https://marsdata.ai/research/reminders-that-start-the-work