The name in the window is a decision, not a default
Whatever appears at the top of a chat window is the first thing a visitor learns about who they are talking to, and most teams never actually decide it. It ends up being whatever was typed into an account form on the day somebody was added, which is why so many chat windows are headed by a full legal name, an internal username, or nothing at all.
It is worth ten minutes of thought, because the name does real work. It sets the register of the conversation, it tells the visitor whether they are talking to a person or a company, and it determines what they can reasonably expect when they come back next week and ask for the same individual.
There is no universally correct answer. There is a correct answer for your team, and it follows from what you are trying to signal and what you can sustain.
Real names, first names, or a team identity
Full real names signal accountability. They suit professional services, high-value sales and any situation where the relationship is with a person rather than a brand. They also raise the stakes: a named individual is searchable, and some of your team will reasonably not want their surname attached to hundreds of public conversations.
First names only are the common middle ground, and for most teams they are the right call. They are human enough to feel like a conversation with somebody rather than something, and impersonal enough to protect the individual. They also survive staff changes gracefully, which full names do not.
A shared team identity — the support team, or simply the company name — suits high-volume operations where conversations are frequently handed between people and no single agent owns the outcome. It is honest about how the work actually happens. Its cost is warmth: a visitor who is having a difficult conversation generally responds better to a person than to a department, and the team identity gives them nobody to thank.
When a pseudonym is legitimate
Consistent working names are common in support, and they are not inherently dishonest. Where the person on the other end is genuinely a person, giving that person a stable working name is a reasonable safety measure — particularly for agents who have been harassed, who share a common name with someone else in the team, or whose own name is repeatedly misread in a way that derails the conversation.
The line worth holding is that a pseudonym is a privacy measure, not a marketing device. Using a working name so an agent is not findable on social media is legitimate. Inventing a roster of friendly-sounding personalities to make a two-person company look like a department is a different thing, and it is the kind of decision that reads badly if a customer ever works it out.
The practical test: if a customer learned exactly how the naming works, would they feel protected, indifferent, or deceived? The first two are fine. If the answer is the third, pick something else.
Photos and avatars
A photograph raises everything the name does, in both directions. A real face makes a conversation noticeably warmer and it is the single cheapest way to make a small team feel present rather than automated. It also exposes the individual more than a name does, and it dates — a photo from four years ago on a person who left two years ago is a small dishonesty running all day.
Avoid stock photography of people who do not work for you. It is the one option here that is straightforwardly misleading, it is recognised more often than teams assume, and the trust it costs when spotted is worth far more than the warmth it borrowed.
A neutral illustration or a company mark is a perfectly respectable middle option. It says nothing false, it never goes stale, and it pairs well with a first-name approach: a human name and a plain avatar reads as a real person who would rather not publish their face, which is a position visitors understand.
Consistency matters more than the choice
Whatever you choose matters less than applying it uniformly. A team where two agents use full names, one uses a nickname and one appears as the company reads as disorganised, and it makes the sensible follow-up question — can I speak to the same person as last time — impossible for a visitor to ask.
Write the rule down with the rest of your voice guidance, including what happens when somebody joins and when somebody leaves. Departures are the case teams handle worst: a returning customer asking for an agent who left six months ago should get a straight answer, not a conversation with a stranger using the same name.
Revisit it when the team changes shape rather than when someone complains. The naming convention that suits three people who know every customer is rarely the one that suits fifteen who do not.
Be clear when it is not a person at all
The one place where this stops being a matter of taste is the boundary between a person and an automated layer. A human name on an automated responder is the version of this decision that customers genuinely object to, because it invites them to attribute intent and judgement to something that has neither.
Give an AI layer a name that does not imply a person, and say plainly what it is when a conversation begins. Our guide to consent and disclosure covers how to word that; the naming point is narrower and simpler, which is that the name itself should not be doing the deceiving.
The reverse case deserves the same care. When a real person takes over from an automated layer, the change of name is the clearest signal available that something has changed, and it is worth making that transition visible rather than seamless.
How MyLiveChat fits
Each agent account carries a display name separate from the login, so what a visitor sees is a deliberate choice rather than whatever was typed into a form. Changing it does not affect the account, which means a first-name convention or a consistent working name is a settings decision rather than an administrative one.
Agent photographs are selected per agent from the same place, so a team can run real faces, a shared mark or a mix without anybody editing markup.
The rest of the impression — the header, the colours, the greeting that arrives before any agent does — comes from your widget customisation settings, and it is worth setting the name and the surrounding presentation at the same time. They are read together, and a warm first name under a header that still says Live Support Window is a mixed signal.
What to measure
This is not a metric-driven decision, and inventing a number for it would be worse than useless. What you can do is read transcripts for the tell: visitors who use an agent’s name back are engaging with a person, and visitors who write to the company in the third person have understood themselves to be talking to an organisation. Either can be correct; the question is whether it matches what you intended.
Watch how returning customers open a conversation. If they routinely ask for someone by name, your naming convention has created an expectation of continuity, and you need a real answer for the times you cannot meet it.
And check the small things once a quarter: that everyone follows the convention, that no photograph belongs to somebody who has left, and that nothing in the window implies a person where there is not one.