Visibility, that is, seeing and being seen, matters at work. In a physical workplace, people can usually see who is present, who is paying attention, where colleagues are looking, and how they respond to what is being said. In digital work, many of these cues disappear. Organisations have tried to compensate by making other kinds of information visible.

There are two known concepts of visibility. One is what researchers tend to refer to as behavioural visibility. Calendars, online status indicators, and other digital traces show what employees are doing, or at least suggest that they are active. Another is communicative visibility. Messages, posts, and other online exchanges allow colleagues to see what people are saying and who is communicating with whom.

These forms of visibility tell us what workers do and say. They do not, however, fully capture how people appear to one another. This is where avatars introduce something new.

A third kind of visibility

What distinguishes avatars is that they act as visible stand-ins for people and give users some control over how they are represented. Based on an analysis of 54 studies, we were able to differentiate how people use avatars to shape what others can perceive about them in digital environments. We call this expressive visibility.

We identified five ways in which avatars create expressive visibility.

1. Avatars convey presence. Even a simple, non-realistic avatar can indicate that a colleague is available or occupying the same digital space. More realistic avatars, supported by a more immersive environment, can create a stronger feeling of actually being together.
2. Avatars support non-verbal communication. A nod, gesture, facial expression, change in posture, or direction of gaze can signal attention and emotion. Relative to audio-only call, such cues from the use of avatars can help people manage turn-taking and interpret how others are responding.
3. Avatars can direct attention to objects. An avatar might point towards a diagram, turn towards a virtual display, or use visible hands to interact with an object. This object referencing can make collaboration in shared digital spaces easier.
4. Avatars make movement visible. Moving closer to someone can signal a wish to join a conversation, while moving away can create distance. An avatar’s location can also help others navigate a virtual environment and support orientation.
5. Avatars communicate identity. Clothing, appearance, and other features can reveal something about the person represented. At the same time, avatars can conceal age, gender, ethnicity, disability, organisational status, or other personal characteristics. This duality exists due to typical customisation features that allow different persona-making which allow expressivity such as true persona, enhanced persona, or false persona.

Where avatar visibility is more or less useful

This final point reveals the central tension in avatar use. Avatars can make workers more visible, but they can also allow people to hide or alter aspects of themselves. This is not necessarily a problem. Sometimes limiting visibility can be useful.

During brainstorming, for example, reducing cues about organisational rank may help participants evaluate ideas on their merits rather than according to who proposed them. In the early stages of recruitment, standardised avatars could conceal demographic characteristics and reduce some forms of bias. However, avatar design may introduce new biases, such as preferences based on the attractiveness or professionalism of the digital figure.

In client meetings, negotiations, or professional consultations, concealment may be much less helpful. These interactions often depend on knowing who one is dealing with and interpreting subtle facial and bodily cues. An avatar that works well in an internal creativity session may therefore undermine trust in a high-stakes client conversation.

Whether an avatar is useful depends on four conditions: what alternatives are available, how much control the user has, how appropriate and intelligible the representation is, and how immersive the environment feels. If live video is available, a basic avatar may appear unnecessary. If users cannot customise their avatar, they may feel misrepresented. If gestures are poorly rendered, colleagues may misunderstand them. And in a simple videoconferencing window, an avatar cannot use space and movement in the same way as in an immersive virtual environment.

The next challenge: Who controls the avatar?

AI-controlled avatars, which are increasingly realistic and referred to as “digital humans,” make these questions even more urgent. Some avatars, including Zoom’s meeting avatars, are animated by a person’s live movements. Others can already deliver a script in a user’s likeness and voice without that person appearing on camera. Increasingly, an avatar may be controlled by a person, by an AI system, or by some combination of the two. It might eventually attend a meeting, answer routine questions, or act on someone’s behalf.

As these uses expand, seeing an avatar may no longer tell us whether the person it represents is actually present. Organisations will need clear rules about when AI is acting through an avatar and how this is disclosed. Otherwise, technologies intended to improve visibility may instead create uncertainty about identity, agency, and authenticity.

Avatars should therefore not be treated as universally better or worse than video, text, or face-to-face interaction. They are tools for managing visibility. Their value lies not only in what they allow people to show, but also in what they allow people to withhold. The important question is not simply whether workplaces should use avatars, but what kind of visibility a particular task requires.