- On-call duty is not free time: on days with on-call duty, mood in the morning was worse, and the cortisol awakening response also changed (Dettmers et al. 2016, 132 people).
- Only working in your free time disrupts switching off: in a diary study, the urge to reply immediately had an effect only through actual work on the phone (Cambier et al. 2019).
- Bundled notifications do better than constant ones: bundling 3 times a day brought more attention and less stress. With no notifications at all, the worry of missing something rose (Fitz et al. 2019, 237 people).
- The right boundary is individual: separating work and private life is not better in itself than blending them. What counts is whether it fits your own preference (Kreiner 2006).
- A concrete plan calms the mind: those who decide when and how an open task will be done are caught up by it less often (Masicampo and Baumeister 2011).
- The body of studies has gaps: almost only employees were studied, mostly by self-report, and all the data comes from the time before ChatGPT.
Employees, at best, have a rule that says whether the work phone still rings after 19 h. Self-employed people and entrepreneurs draw this line themselves. That is why it is worth looking at the research on constant availability, because it distinguishes more precisely than gut feeling. This post sorts 15 studies into 4 topics: availability after hours, notifications, boundaries between work and private life, and planning. At the end comes an assessment for working with AI, clearly separated from the findings. If you are currently introducing AI in your company, you'll find the basis here for rules that apply alongside the tools.
Constant availability after work and switching off
The clearest finding comes from a study that compared the same people on days with and without on-call duty. In 2016, Dettmers and colleagues followed a total of 132 people for 4 days with and 4 days without on-call duty, and for 51 of them a cortisol measurement in saliva was added. With on-call duty, mood in the morning was worse, and the cortisol awakening response also turned out differently. The authors explain the worse mood by the lack of control over one's own free time; for cortisol they did not find this link. Their conclusion: hours on call cannot be counted as free time.
2 diary studies look more closely at the phone. In Derks and colleagues (2014, 70 employees, 268 recorded days), more work on the smartphone outside working hours went along with less detachment and, through that, more exhaustion. The effect was especially strong for people whose workplace expects a separation between work and private life. Cambier and colleagues (2019, 80 employees) separated the urge to respond from actually responding. The urge alone did not directly disturb detachment, only the work on the phone did.
A meta-analysis summarizes this more broadly. In 2017, Wendsche and Lohmann-Haislah analyzed data from 86 publications with 38,124 people. Mentally switching off went along with less exhaustion, better sleep, higher life satisfaction and somewhat better performance, while work in free time clearly went along with less switching off. They found no association with physiological stress measures, and even a negative one with creativity.
Even so, the picture is not entirely clear. A review of 78 studies (Schoellbauer et al. 2023) describes the results on work outside working hours as inconsistent. The authors distinguish: frequent, short work outside working hours is a strain, while longer continuous blocks can help. Private time is lost in both cases.
For the self-employed this means: a planned work block in the evening is something different from constant availability in which a message is answered every 20 minutes. One question that concerns many remains open: none of the studies examined whether merely seeing business appointments on your private phone already disrupts switching off.
Bundle notifications instead of turning them off
Even a single notification costs concentration. In a lab experiment by Stothart and colleagues (2015), participants received a call or a text message during an attention task, which they did not respond to. The mere notification worsened performance about as much as actively using the phone.
The obvious conclusion would be to turn everything off. A field experiment argues against that. In 2019, Fitz and colleagues divided a total of 237 people into 3 groups: notifications as before, bundled 3 times a day, or none at all. The bundled group reported more attention, better mood, more control over their own phone and less stress. The group without notifications had hardly any benefits, but more anxiety and more worry about missing something. A subordinate clause in the abstract also mentions lower productivity for the bundled group, which contradicts the other findings. I did not check the full text on this.
For e-mails, Kushlev and Dunn (2015) arrive at a similar result. 124 adults checked e-mail only 3 times a day for one week and without limit for another week. In the limited week they had less daily stress.
Scheduled summary on the iPhone
iOS bundles notifications with the Scheduled Summary: turn on Settings > Notifications > Scheduled Summary, set the times, and under "A to Z" choose the apps whose notifications should arrive together. For each app, you can choose between "Immediate Delivery" and "Scheduled Summary" under "Notification Delivery" (Apple Support).
Phone and calendar reminders should arrive immediately, otherwise the reminder of an appointment may only show up afterwards. For mail, messengers and project tools, bundled delivery fits. The studies worked with 3 times a day. Whether 2 times work just as well was not examined in these studies.
Android and Google Pixel
For Pixel devices, the Google help pages do not describe a scheduled summary like on the iPhone. There are modes such as “Do Not Disturb” (can be scheduled, but only mute), snoozing individual notifications, from Android 14 the “Reduce repeated notifications” feature, and AI summaries of notifications. On a Pixel, bundling therefore works like this: have business apps delivered without sound and vibration and check them yourself at fixed times, while phone and calendar stay loud. Devices from other manufacturers sometimes have their own features.
Boundaries between work and private life
For a long time, the recommendation was to merge work and private life as much as possible. In 2006, Kreiner questioned this with a survey of 325 employees: conflict, stress and satisfaction are related to fit, that is, whether the actual separation matches what a person wants. Rothbard, Phillips and Dumas (2005, 460 employees) showed the same from another angle. People who wanted separation were less satisfied and less committed to the company with integrating offerings such as on-site childcare, but more committed with separating offerings such as flextime. More offerings are not automatically better. A third survey (Park, Fritz and Jex 2011) found that the wish for separation and the norm of separation are linked to better detachment, partly because less work technology is used at home.
Kreiner, Hollensbe and Sheep (2009) describe how people keep their boundaries in 2 qualitative studies. They found 4 types of tactics: behavioral, temporal, physical and communicative. A boundary violation here means anything that breaks the boundary a person wants for themselves, not every kind of stress.
For the self-employed, the 4 tactics can be translated directly. The following examples are practical recommendations derived from the study, but not studied specifically:
- Behavior: separate numbers or devices for work and private life, business calendars on your personal phone at most read-only and without reminders.
- Time: fixed time windows in which inquiries are answered, with a buffer for urgent matters.
- Space: Work devices stay at the workplace, even if it is in your own home.
- Communication: a commitment such as “reply within one working day”, stated openly in your signature, on your website and in the introductory call.
The last point deserves weight. Self-employed people limit constant availability themselves, and a stated commitment creates reliability without requiring constant presence. Anyone who knows by when a reply will come is probably less likely to write a second time.
Planning with a buffer
Unfinished tasks make themselves heard. In 2011, Masicampo and Baumeister showed in lab experiments that unfinished goals push into your thoughts and worsen performance on other tasks. A concrete plan for when and how the goal will be completed ends this pressure, even though the task is still open. For the end of the workday, a transfer suggests itself, but it has not been studied: if you note the open items with a when and a how at the end of work, you are likely to carry them home in your head less often.
Plans do need some slack, though. In 2018, Parke and colleagues analyzed daily surveys over 2 weeks. Planning went along with more engagement and better daily performance, but on days with many interruptions the effect weakened. People who planned for interruptions in advance, that is, thought about how to handle them, benefited even on turbulent days. Applied to the calendar, this is an inference: a day plan without free time for calls and follow-up questions is likely to hold up less well than one with a buffer.
A third study concerns the household, not the company. In 2019, Daminger conducted 70 interviews in total with 35 couples about cognitive work in the household: anticipating, searching for options, deciding, monitoring. This work is burdensome, often invisible and a frequent source of conflict, and women take on more of it. Applied to the self-employed, this is an interpretation, not a finding: in a company too, keeping open items in your head is work that appears in no calendar.
The 15 studies at a glance
All details come from the abstracts, meaning the summaries of the studies. I did not read the full texts. The PubMed links show the abstract for free, the DOI links lead to the publisher.
| Study | Studied | Result |
|---|---|---|
| Kreiner 2006 | Survey, 325 employees: desire for separation and actual separation | The fit is linked to conflict, stress and satisfaction. |
| Rothbard, Phillips, Dumas 2005 | Survey, 460 employees: blending and separating offerings | People who want separation are less satisfied and less committed with offers that mix things together, and more committed with offers that separate them. |
| Kreiner, Hollensbe, Sheep 2009 | 2 qualitative studies: tactics for your own boundary | 4 types of tactics: behavior, time, space, communication. |
| Park, Fritz, Jex 2011 | Survey of employees: separation, norms, technology at home | The wish for separation and the norm of separation are linked to better detachment. |
| Wendsche, Lohmann-Haislah 2017 | Meta-analysis, 86 publications, 38,124 people | Detaching goes along with less exhaustion and better sleep, work in free time with less detaching. |
| Schoellbauer et al. 2023 | Overview of 78 studies from 2007 to 2021 | Mixed: frequent work outside of working hours is a burden, longer sessions can help. |
| Dettmers et al. 2016 | 132 people, 4 days each with and without on-call duty, cortisol at 51 | With on-call duty: worse morning mood and altered cortisol awakening response. |
| Derks, van Mierlo, Schmitz 2014 | Diary study, 70 employees, 268 days | More work on the smartphone outside working hours, less detachment, more exhaustion. |
| Cambier, Derks, Vlerick 2019 | Diary study, 80 employees: urge to reply | Only actual work on the phone disrupts switching off. |
| Stothart, Mitchum, Yehnert 2015 | Lab experiment: phone call or text message during a task | The mere notification worsens performance. |
| Fitz et al. 2019 | Field experiment, 237 people: as before, bundled, none | Bundled is better than constant, without any more fear of missing something. |
| Kushlev, Dunn 2015 | Experiment, 124 adults: e-mail 3 times a day or unlimited | Less daily stress in a limited week. |
| Parke et al. 2018 | Daily surveys over 2 weeks: planning and interruptions | Planning helps, but with many interruptions only if they are planned in advance. |
| Masicampo, Baumeister 2011 | Lab experiments: thoughts about unfinished goals | A concrete plan ends the mental nagging. |
| Daminger 2019 | 70 interviews with 35 couples: thinking work in the household | Thinking ahead and monitoring are taxing and often invisible. |
Gaps in the evidence and context for working with AI
Gaps in the evidence
The evidence has clear limits. Almost only employees were studied, most studies rely on self-report and show associations, not causes. Experiments are the exception (Stothart, Fitz, Kushlev, Masicampo), plus a comparison of the same people on days with and without on-call duty (Dettmers). I found no suitable study on solo self-employed people and constant availability. Applying it to self-employed people and entrepreneurs is therefore an inference, even though the mechanisms, such as lack of control over free time, should apply to them just as well.
Context: faster pace of work since ChatGPT
This section is an assessment, not a study finding. All 15 studies are based on data from the time before ChatGPT, which OpenAI released at the end of November 2022. The 2023 overview also evaluates studies up to 2021.
Since then, the pace of work in many companies has probably changed. AI tools deliver drafts, evaluations and research in minutes, and AI agents handle individual steps independently. As a result, more interim results, follow-up questions and approvals arrive in a shorter time. I have not found a study on whether this amplifies the effects of constant availability. My take: the rules from the studies are likely to become more important with AI. An agent can report its results in batches just like a mailbox, and approvals can be placed into fixed time windows instead of being spread across the day.
I show what workflows like these look like in everyday work every Sunday in the AI Sync newsletter.
Everything on this page can be implemented without me. I add new tools and changes here, and every Sunday there is a tip about it in the newsletter.
To AI SyncI set up AI processes with you that deliver their results collected at fixed times, and we define together who approves. On site from Eisenstadt to Mattersburg and Wiener Neustadt, online across the whole German-speaking region.
Learn moreFrequently asked questions
How does constant availability affect recovery?
In a study with 132 people, mood in the morning was worse on days with on-call duty, and the cortisol awakening response also changed (Dettmers et al. 2016). A meta-analysis with 38,124 people links work in free time with less mental switching off, and less switching off with more exhaustion and worse sleep (Wendsche and Lohmann-Haislah 2017).
Should self-employed people turn off all notifications?
The studies favor bundling. In the field experiment by Fitz et al. (2019) with 237 people, notifications bundled 3 times a day brought more attention and less stress. With no notifications at all, there were hardly any benefits, but more worry about missing something. Calls and calendar reminders should still arrive immediately.
How often a day should you check e-mail?
None of the studies determined an ideal number. In the experiment by Kushlev and Dunn (2015), 124 adults checked their e-mails 3 times a day for a week and had less daily stress that week than in the week without limits. Fixed times in the morning, at noon and in the afternoon are an obvious starting point.
How can self-employed people limit their availability?
Kreiner and colleagues (2009) describe 4 types of tactics: behavioral, temporal, physical and communicative. For self-employed people they can be put into practice like this, even though these examples have not been specifically studied: separate numbers or devices, fixed reply windows, work devices kept at the workplace, and an openly stated commitment such as "reply within one working day". According to Kreiner (2006), what matters is that the boundary fits your own preference.
Do the studies also apply to working with AI?
The data comes from the time before ChatGPT, the end of November 2022. I did not find a study on whether faster processes with AI amplify the findings. But the mechanisms, such as interruption by notifications and lack of control over free time, should apply regardless of the tool. Bundled notifications and fixed approval times can also be set up for AI processes.
Changes to this post
- 21 October 2026First version.
