The Zendesk Chat Dashboard

Live monitoring and reporting are two different jobs, and a Zendesk chat dashboard gets used for both. Confusing them is why chat dashboards go unread.

Start free trial All guides

Two things called a Zendesk chat dashboard

The first is operational. Who's online, how many conversations are open right now, who's waiting and for how long. You look at it constantly during a shift and never afterwards.

The second is analytical, and it lives in Explore. Volume by hour, first reply time, satisfaction, missed conversations, agent throughput over weeks. You look at it on a Monday and act on it once a month.

Trying to make one screen do both produces a screen nobody uses. Build them separately. The broader argument is in Zendesk dashboards, and it applies doubly to chat, because chat is the channel where a five minute delay isn't a slow reply, it's a lost customer.

Which live view your plan includes has varied over the years, and the legacy Chat dashboard was a different screen from the monitoring in the agent workspace. Check the current Zendesk docs against your account rather than following a screenshot from a forum post.

The live board, and what belongs on it

Four numbers. If a team lead has to interpret it, it's a report, not a monitor.

Agents online versus expected. The single most common cause of a bad chat afternoon is that two people forgot to change status after lunch.

Conversations waiting, and the oldest wait time. The oldest one is the number that matters. An average of forty seconds can comfortably hide somebody sitting at nine minutes.

Open conversations per agent. Spot the person quietly carrying five while somebody else has one, before they burn out or start rushing.

Missed or abandoned so far today. Cumulative, visible, uncomfortable. That's the point of it.

Put it on a screen the team can see, or pin it in a shared browser tab. A monitoring board behind a login is a board nobody opens.

What does not belong on it: anything historical. Yesterday's satisfaction score changes nothing about the person waiting right now, and every extra tile makes the four that matter harder to read at a glance from across the room.

The weekly numbers worth reporting

Different set entirely, and a much shorter list than most teams build.

First reply time, at the median and the 90th percentile. Averages on chat are close to useless, since one abandoned conversation with a 45 minute reply drags the whole thing sideways.

Missed conversation rate by hour of day. This is the one that changes a rota. If you're missing 30% of chats between 4pm and 6pm, you don't have a chat problem, you have a staffing problem with a very specific shape.

Volume by entry page. Tells you which page is generating questions, which is a product or content signal rather than a support one. Take it to the people who own that page.

Satisfaction, read with suspicion. Chat survey response rates skew hard toward people who had strong feelings either way, so treat the trend as the signal and the absolute number as decoration.

Four numbers, one page, twelve weeks of history. If somebody asks for a fifth, ask what they would do differently depending on the answer, and watch how often there isn't one.

Where it falls short

Three honest limitations, and knowing them saves you from arguing with a number in a meeting.

Attribution across channels is weak. A conversation that starts in chat, gets escalated, and finishes over email reports as a chat with an odd resolution time. You're looking at a fragment of a customer journey and calling it a chat metric.

Concurrency distorts handle time. An agent with three conversations open has a handle time reflecting waiting, not working. Comparing chat handle time to email handle time is meaningless, and people do it constantly, usually in a slide.

And bot-handled conversations muddy every count, depending on how your account attributes them. Before you celebrate a drop in agent conversations, check whether those conversations were resolved or merely absorbed. Those aren't the same thing, and only one of them is good news.

One filter worth adding to everything

If chat conversations become tickets in your account, exclude anything tagged closed_by_merge from volume and handle time widgets.

Merged duplicates are administrative closures rather than work. Leaving them in inflates your conversation count, flatters your solved-per-agent number and drags the median handle time down, which makes the whole board slightly wrong in a way that's genuinely hard to spot.

FAQ

Frequently asked questions

Where does chat reporting actually live?

In Explore, not in the live view. Zendesk chat reporting on volume, wait time and satisfaction is built there, and Zendesk chat analytics beyond the prebuilt dashboard needs the paid Explore tier.

Does Zendesk have a live chat monitoring view?

Yes, an operational view of who is online and what is waiting, separate from the historical reporting in Explore. Which view your plan includes varies, so check current Zendesk docs.

Which chat metric matters most?

Missed conversation rate broken down by hour. It is the only one that reliably tells you to change something.

Why is chat handle time so much lower than email?

Because concurrency means an agent is waiting rather than working for much of it. The two numbers are not comparable and should not sit on the same chart.

Can I email a Zendesk chat dashboard on a schedule?

Explore dashboards can be scheduled. Keep the live monitoring board live and schedule the weekly trend, because scheduled numbers actually get read.