A person opens a public-service app to book an appointment. The task takes four minutes, the instructions are clear and the confirmation can be saved. Later, the same person opens another app to message a friend and spends far longer than intended following unrelated recommendations. Both experiences appear in a daily screen-time total. Digital wellbeing asks a wider question: whether digital tools support a person’s health, relationships, learning and chosen purposes without taking away their ability to understand and direct their own use.
The term has no single universally accepted score. It can refer to subjective experience, patterns of use, access, safety, attention or the design of institutions and platforms. A useful definition must say what is being evaluated. A person may benefit from long hours online because that is where work, disability access or family contact happens. Another may feel depleted after a short but hostile interaction. The issue is not whether a screen is present. It is how a technology, a task and a life context meet.
Why duration cannot define the whole construct
The OECD’s 2025 report on screen time and subjective wellbeing analyzed personal screen use across fourteen countries and found that people reporting more than five hours per day had higher odds of several poor wellbeing outcomes than a moderate-use group. Very low use was also associated with some poorer outcomes. The report is important because it provides standardized comparisons and includes life satisfaction and sense of purpose as well as mental wellbeing. But its cross-sectional associations cannot prove that time on screens caused the outcomes.
A duration figure combines different purposes. A medical appointment, language practice, a game with friends, paid work, harassment and endless recommendation loops can all contribute minutes. These activities differ in control, benefit, burden and exposure to risk. Even within one activity, a person may start with one intention and be diverted. Digital wellbeing therefore needs measures of task completion, agency, safety and subjective experience alongside time, rather than a single threshold presented as a universal rule.
Orben and Przybylski’s large-scale analysis of adolescents found a small average association between digital technology use and wellbeing in their datasets. Their work cautions against inflating a broad usage measure into a complete explanation of young people’s lives. A small average does not mean no subgroup or use pattern matters. It means that research must specify who is using what, how and under which circumstances. The same care applies to adults and older users, whose digital activities and opportunities differ.
The Institute treats digital wellbeing as a relation between people and the systems they are asked to use. This is editorial synthesis, not a validated diagnostic construct. The definition includes what the person can achieve, what costs the technology imposes and whether they can leave, repair an error or choose a different route. A design that maximizes engagement can undermine wellbeing even if users report a moment of pleasure; a necessary service can support wellbeing even if it is not entertaining.
Experience, access and agency
Subjective wellbeing asks how life is going for the person. Digital wellbeing adds questions about whether a technology helps or frustrates that life. Can a learner find trustworthy material? Can a worker complete a task without avoidable interruptions? Can a family member communicate across distance? Can someone with a disability use the service without depending on a workaround? These are concrete outcomes, not just feelings about a device.
Access is part of the picture. Being offline by choice may be restorative. Being offline because a service is inaccessible, unaffordable or unavailable can exclude a person from care, work and civic participation. A “less screen time” program that assumes everyone can reach a physical alternative may widen inequality. Digital wellbeing must include meaningful options: usable online routes, human alternatives where needed and the freedom to choose among them.
Agency concerns the ability to set a purpose and act on it. A person may decide to read, contact someone or finish a form; the technology may help, interrupt or redirect them. Clear navigation, understandable defaults, visible exits and repairable errors support agency. Notifications, deceptive timing or hidden subscriptions can work against it. Measuring agency requires observing task paths and asking users what they intended, not simply counting minutes or clicks.
Safety matters independently of average satisfaction. Harassment, fraud, privacy violations and exploitation can make a digital environment harmful. A user may still rely on that environment for essential contact or income, which makes “just leave the app” an inadequate response. Institutions and platform designers have responsibilities to reduce hazards and provide recourse. Digital wellbeing should not become a way to shift every burden onto the individual user.
Connection is not the same as constant contact
Digital communication can maintain relationships across countries and schedules. It can also create pressure to remain available, compare oneself with others or manage multiple shallow exchanges. A count of messages tells little about trust or support. The WHO’s social-connection framework distinguishes social isolation from loneliness; a person can be frequently contacted yet feel alone. The digital question is whether communication serves the relationship the people want, not whether the platform reports a high engagement rate.
For a group, design choices matter. A school might use one channel for essential updates and leave students free from late-night alerts. A workplace might state response expectations and allow focus periods. A family might agree on when a call is welcome rather than expecting immediate replies at all hours. These practices are plausible ways to protect attention and connection, but they must be evaluated locally rather than claimed as universal treatments.
Digital contact can also be a first step toward participation. Someone may find a local group online before feeling able to attend in person. Another may prefer an online group because physical spaces are unsafe or inaccessible. A policy that treats online ties as inherently inferior would miss those realities. A better evaluation asks whether people feel more supported, can seek help when needed and can choose the mode that fits their circumstances.
Learning, work and the design of attention
Technology can broaden access to information and feedback. A learner can replay an explanation, collaborate across distance or use assistive tools. It can also make study fragmented when every page competes with alerts, recommendations and separate login systems. The educational outcome is not how long a learner was online. It is whether they can understand, remember and apply the material, and whether the environment respects their attention and privacy.
For workers, digital systems can remove tedious steps or create new ones. A tool that automates a form may save time while making errors difficult to contest. A dashboard may make a process visible while also inviting constant monitoring. Wellbeing evaluation should include workload, autonomy, error correction and the distribution of benefits. A company cannot infer that employees are well because average task duration fell; the saved time may be filled with new demands.
Workplace technologies also redistribute responsibility. If a scheduling system creates a clash, can a worker identify the rule that produced it and ask for a correction? If a message platform makes everyone appear continually available, has the organization agreed on response times? Individual settings can help, but they cannot override a manager’s expectations. A digital-wellbeing assessment should therefore inspect norms and authority alongside interface design. Otherwise it may recommend personal discipline to someone whose task requires the very behavior being criticized.
The distribution of effects is another test. A digital service may reduce average waiting time while increasing the burden for users who need translation, assistive technology or a human explanation. A school platform may work well for students with quiet rooms and reliable devices while creating extra work for those without them. Reporting only a mean satisfaction score can hide these differences. A useful evaluation asks who completes the task, who leaves it unfinished and how the service responds when a person cannot follow the expected path.
The OECD Digital Well-being Hub’s work asks how digital transformation relates to lived experience. The useful implication is to pair usage indicators with wellbeing and social measures rather than assuming technology adoption is itself the goal. A faster service can be good for a person if it is usable and accountable. If it excludes someone who needs help, the speed statistic conceals harm. Digital wellbeing requires attention to both the average and the people left out.
The same logic applies to generative AI. A system can help a person draft, translate or explore an idea while making it harder to distinguish evidence from plausible text. The question is whether the person can inspect sources, revise an output and retain responsibility for a consequential choice. Enjoyment or efficiency alone is not enough. An AI-assisted workflow should be tested for understanding, error detection and the ability to continue when the system is unavailable.
How to study the concept without making it vague
A study or service review should start with a named activity and outcome. For a booking app, the outcome might be successful completion, comprehension of the appointment details, privacy and access to correction. For a social platform, it might be the quality of desired interaction, unwanted exposure and the ability to stop. For an educational tool, it might be learning that transfers to a new problem. “Improved digital wellbeing” should be a conclusion from these measures, not a replacement for them.
Mixed methods can help. Logs may show timing and task paths; surveys may capture felt control and wellbeing; interviews may reveal barriers that neither measure sees. Each method has limits. A log cannot know why someone paused, a questionnaire can invite socially desirable answers and an interview may not represent a whole population. Combining them carefully can show where they agree and where a new question is needed.
Longitudinal data improve on a one-time snapshot by observing change, but they do not automatically establish cause. A person may change their digital habits because of a new job, illness or relationship. An intervention study can test a specified design change against a credible comparison, but its effect may depend on user group and context. Strong work states what changed, who had a choice, what outcome was measured and how long the result lasted.
Privacy is a methodological and ethical condition. Digital traces can expose health, relationships, location and beliefs. A project should collect only what is needed for a clear question, protect participants and avoid turning wellbeing research into surveillance. A user who changes behavior because every action is tracked is not experiencing the same environment as an ordinary user. The measurement method can alter the phenomenon.
Involving users in evaluation can reveal measures that experts overlook. A person may value being able to pause a task and return without losing progress; another may value a clear record of who saw their information. These are not minor preferences if they determine whether a service is usable under real conditions. Participatory feedback does not replace outcome data, but it can identify the outcomes worth testing. The designer should then check whether the proposed improvement works across users, not only among those who were easiest to recruit.
What the evidence does not show
The OECD’s cross-country screen-time associations do not show that one duration limit will make everyone healthier. Adolescent technology-use correlations do not prove that digital tools as a category cause poor mental health. Nor does a person’s satisfaction with an app prove that the app supports their long-term goals, privacy or ability to leave. These outcomes require separate observation and, where possible, causal tests of specific design changes.
Digital wellbeing is not a diagnosis and cannot be read from a device dashboard. A person with heavy use may be working, learning or maintaining family contact; a person with little use may be excluded. The construct should not be used to blame users for platform design, employer demands or inaccessible public services. It also should not be used to promise that a short “detox” will resolve distress with other causes.
The evidence does not support declaring all online connection shallow or every automated tool dehumanizing. Neither does it support an assumption that more technology is always progress. Each claim needs a defined task, comparison, outcome and population. A good assessment can identify both useful affordances and costs without demanding one verdict on “the digital world.”
Building a more useful question
Return to the two apps in the opening scene. The appointment app supports a chosen task and leaves a clear record. The messaging app may support a friendship, while the recommendation path may redirect attention. A useful review asks which design choices made these experiences different, which benefits the user wanted and which costs they would choose to change. The same person can value a technology and criticize its design.
Digital wellbeing, used precisely, is not a contest to minimize minutes. It is the quality of a person’s digital experience within the conditions they inhabit. That quality depends on access, purpose, safety, attention and recourse. Research can measure pieces of it, and institutions can improve those pieces. The term earns its place when it makes the pieces visible rather than hiding them inside a reassuring score.
That makes it a continuing design responsibility.
Research references
- OECD. (2025). Screen time and subjective well-being: Insights from a few countries worldwide. View source
- Orben, A., & Przybylski, A. K. (2019). The association between adolescent well-being and digital technology use. Nature Human Behaviour, 3, 173–182. View source
- Przybylski, A. K., & Weinstein, N. (2017). A large-scale test of the Goldilocks hypothesis: Quantifying the relations between digital-screen use and the mental well-being of adolescents. Psychological Science, 28(2), 204–215. View source
- Huang, S., Lai, X., Li, Y., Cui, Y., & Wang, Y. (2023). Beyond screen time: The different longitudinal relations between adolescents’ smartphone use content and their mental health. Children, 10(5), 770. View source
- World Health Organization. (2025). From loneliness to social connection: Charting a path to healthier societies. View source
- Holt-Lunstad, J. (2022). Social connection as a public health issue: The evidence and a systemic framework for prioritizing the social in social determinants of health. Annual Review of Public Health, 43, 193–213. View source