The hardest conversations in a busy inbox are not the ones that are loud. They are the ones that are quietly going wrong while the team works through the queue in order. HellouOne's new customer sentiment feature is built for the supervisor who needs to find those conversations in seconds. It shows a mood dot on every conversation card, adds mood and priority chips, and gives sentiment its own page under Reports.
Where you will see it
A mood dot on every conversation card. Scanning the inbox list, you can see at a glance which conversations are calm and which are not, without opening any of them.
Mood and priority chips. Inside a conversation, chips show the current mood and the priority, so whoever picks it up knows how to open.
Reports > Sentiment. A dedicated report that shows how customers feel over time, so a change in mood across a week stands out from a single bad day.
The mood comes from HellouOne's AI classifications, the small model that reads every incoming message. Sentiment is one of the things it reads, which means the dot updates as the conversation moves: a customer who started annoyed and ended satisfied shows that journey, and a customer who is getting more frustrated with each reply shows that too.
The dot sits at the top of each conversation card, next to the inbox name; with a single inbox the name does not appear and the dot stands alone. The colors are green for positive, gray for neutral, yellow for negative or annoyed, orange for frustrated, and red for hostile, or for a conversation that requires attention without another reading. When a conversation requires attention, the dot carries a blinking red ring, and hovering over the dot shows the detail, for example hostile, requiring attention, high urgency, and a customer who wants to talk to a person. On the phone there is no hover, so the dot itself carries the signal.
A conversation requires attention when the customer is frustrated or hostile, has high urgency, asked to talk to a person, is at risk of abandonment, or uses offensive language. For a team lead scanning a busy list, the blinking ring is a triage tool: it points at the conversation to pick up first, without opening twenty chats to find the one that is about to go wrong.
Inside a conversation, the sentiment label next to the contact name, such as Hostile or Positive, shows an up arrow when the conversation is improving and a down arrow when it is getting worse. The priority label sits beside it, and you can click it to change priority without leaving the conversation: None, Urgent, High, Medium, or Low. Sentiment updates by itself with each customer message, so nobody on the team scores anything by hand. The arrow matters as much as the label: a hostile conversation with an improving arrow is a recovery in progress, and a neutral one with a worsening arrow is the one to join early, while a small intervention still works.
Why supervisors asked for this
In a team of six agents handling WhatsApp, web chat and email, a supervisor cannot read every thread. What they can do is decide where to spend their attention. Sentiment turns that decision from a guess into a glance. An unhappy customer in a long queue is the one most likely to leave, to post a review, or to escalate, and the one where a quick intervention from a senior person pays off most. With the dot visible on the card, the supervisor jumps to that conversation first, reads the chips, and decides whether to coach the agent, take it over, or let it run.
The priority chip works with the mood. A frustrated customer with a low-stakes question and a calm customer with an urgent one both deserve attention, but of different kinds. Seeing both chips side by side keeps the team from treating every unhappy message as a fire and every polite one as fine.
What the report adds
The inbox view is about now. Reports > Sentiment is about patterns. Over a week or a month, it shows whether customers are, on balance, happier or less happy with the service, and lets the team connect that to what changed: a new AI agent prompt, a product issue, a holiday rush. It is also a fair way to talk about quality with the team, because it reflects what customers expressed rather than what a reviewer thought of a sample of chats.
Sentiment also helps with staffing decisions that are otherwise made on gut feel. If negative mood clusters on Monday mornings or on one channel, that is a scheduling or a content problem you can fix, not a mystery. And because the dot reflects what the customer expressed rather than how long the conversation lasted, a short, sharp complaint gets the attention it deserves instead of hiding behind a low message count.
The report, indicator by indicator
Open Reports, Sentiment and choose the period. Each conversation counts once, by its last reading, no matter how many messages it had. The top indicators are Conversations read; End negative, the percentage of customers who leave dissatisfied; Net sentiment, the percentage ending positive minus the percentage ending negative, from minus 100 to plus 100; Recovered, of the conversations that started negative, how many ended neutral or positive, counting only conversations read at least twice; and Require attention now, the open or pending conversations that need someone right now, independent of the chosen period.
Below that, How conversations end charts the last sentiment of each conversation by day, or by week, month, or year in long periods, so you can see whether this week is going better or worse than the last. The Start to end table shows how conversations started in rows and how they ended in columns, for conversations read at least twice. Green cells improved and red cells worsened, and clicking any number opens those conversations.
The breakdown by Agents, Inboxes, or Teams attributes each conversation to whoever was attending it at its last message, including the AI agent, and conversations nobody attended appear as Unassigned in the Agents view. To compare people fairly, look at the Recovered column and not at how many negative conversations each person received, because whoever handles the difficult cases will always inherit more negatives. Rows marked Few have fewer than 10 conversations, too few to compare.
Frustration by inbox shows the highest frustration each conversation reached (Calm, Annoyed, Frustrated, or Hostile) by inbox, and clicking a segment opens those conversations. Risk signals count how many customers asked to talk to a person, were at risk of abandonment, or used offensive language, plus the median time until an agent responded after someone asked for a person. Sentiment, CSAT and speed shows average CSAT by how the conversation ended, and median first response and resolution by how it started, which helps you see, for example, whether customers who arrive annoyed wait longer than the rest. The Require attention now list closes the report, with a link to open each conversation.
Availability, automation and privacy
Sentiment is available on plans that include AI classifications. The report is for administrators; agents see sentiment in their own conversations. If your plan includes it and you still see nothing, write to support to activate it on your account. With an automation rule you can raise priority or notify a team when a conversation requires attention.
To read sentiment, HellouOne uses the customer’s new messages, the conversation channel, and up to 3 previous messages, including your team’s replies. Before sending them, emails, phones, links, social handles, and document numbers are hidden, and the customer’s name and contact data are never sent.
How to use it in HellouOne
Sentiment is on wherever AI classifications are on, which is from the Growth tier upward and in the free trial. A simple routine works well: at the start of a shift, sort or scan the inbox for negative mood dots and open those first; during the day, use the chips to brief whoever takes over a conversation; at the end of the week, open Reports > Sentiment and look for the days or channels that moved. Pair it with the reply check, which holds risky drafts, so an unhappy customer is not met with a reply that makes things worse. And if a supervisor wants a quick answer, LIA can read the sentiment report and summarise it in the admin chat.
Sentiment is one of the signals supervisors use for the handover described on the AI + human handoff page. Source: Sentiment, how your conversations end. See the WhatsApp team inbox in HellouOne.





















