Color-coded labels with regex and AI-powered auto-tagging, per-label SLAs, routing rules and bulk filters. Stop asking agents to remember to tag — let the rules engine do it the moment the message lands, then report on every label as a first-class metric.
Most teams set up labels in their first week of using a support tool — Refund, Bug, Feature Request, VIP, Billing, Sales — and within a month nobody is applying them consistently. The senior agent tags everything; the juniors forget; the analytics dashboard becomes useless because half the conversations are untagged or wrongly tagged. By month three, leadership is asking "how many refund requests did we get last month?" and the honest answer is "we don't really know — maybe 200, maybe 600".
The second failure mode is labels with no consequence. A conversation gets tagged "VIP" but nothing changes — same queue, same SLA, same agent. Labels become a write-only descriptor instead of a control mechanism. Operators eventually stop trusting the system and rebuild routing logic in their head, defeating the entire point of having tags.
Wachat's chat labels are designed so that tagging happens automatically and labels actively drive behaviour. A label is a first-class object with a colour, an icon, optional SLA overrides, optional routing rules, optional auto-replies, and analytics. The label "Refund" might have a 30-minute SLA, route to the finance team, trigger an internal note to the assignee, and roll up into a weekly executive report — all without an agent ever clicking "Add label".
Wachat labels are workspace-scoped objects with a name, colour (12-step palette), icon, description and parent (for hierarchical labels like "Refund > Damaged" and "Refund > Wrong size"). Any conversation can have any number of labels, applied manually by an agent, automatically by a rule, or by the AI tagger. The label panel in the inbox shows active labels as removable chips with their colours; the queue filter supports any combination of include/exclude labels for precision triage.
Auto-tagging runs on every inbound message. Rules can use exact keyword match, regex (full PCRE), customer attribute conditions (e.g. ltv > 50000), order conditions (e.g. order total > 10k), channel filters, or AI-based intent classification. A typical setup might be: regex /refund|return|money back/i tags "Refund"; regex /not working|broken|error/i tags "Bug Report"; AI intent "complaint" tags "Escalation"; CRM lookup "ltv > 50000" tags "VIP". Multiple rules can apply to one message; the conversation accumulates the union of all matched labels.
Each label can carry its own behaviour. A "VIP" label can override the default SLA from 30 minutes to 10 minutes, force assignment to a senior agent pool, post a Slack alert and pin the conversation to the top of every queue. A "Refund" label can route to the finance team, attach the company's refund policy as an internal note, and require a senior agent's approval before resolution. The behaviour is configured per label in the admin UI — no flow needed for most cases.
Analytics treat every label as a measurable dimension. The labels dashboard shows volume, average response time, average resolution time, CSAT and revenue impact per label per week. Export by label to CSV for board reports, slice by label inside the analytics module, or use labels as audience criteria in Broadcasts ("send to all contacts with the Bug-Report label closed in the last 30 days"). Labels become the spine of the company's operational view of its customer conversations.
Create unlimited labels with name, colour, icon, parent and description. The 12-step palette is colourblind-safe, and labels render as chips in the queue, the conversation header and inside reports for instant visual scan.
PCRE-flavoured regex rules run on inbound message text. Case-insensitive, multi-line and lookahead supported. Test rules against historical conversations in the rule editor before publishing — see how many would have matched in the last 30 days.
A trained classifier maps inbound messages to intents (complaint, refund, sales-enquiry, technical-issue, praise, spam) with confidence scores. High-confidence matches tag automatically; low-confidence ones queue for human review with the suggested label.
Override the team or channel SLA when a label is applied. VIP gets 10 minutes; Refund gets 30 minutes; Bug Report gets 4 hours. The most aggressive applicable SLA wins, so a VIP refund still triggers the 10-minute clock.
When a label is applied, optionally re-assign the conversation to a specific team or agent. A "Refund" label routes to finance; "Billing" routes to accounting; "Bug Report" routes to the on-call engineer. Routing fires once per label application to avoid loops.
Filter the queue by any label combination and apply bulk actions: re-assign, re-label, snooze, send a broadcast template to all linked contacts, or export. Useful for backlog cleanup and for running a campaign targeted at a specific customer state.
Per-label dashboards: volume, response time, resolution time, CSAT, revenue and refund amount linked. Compare labels week over week, segment by channel, or drill into individual conversations from any cell of the dashboard.
A skincare brand has parent label "Refund" with children "Damaged", "Wrong Size", "Allergic Reaction", "Changed Mind", "Late Delivery". Auto-tagging routes "Allergic Reaction" to a senior agent with a 15-minute SLA and a templated apology. Monthly board report shows refund volume by reason without anyone manually classifying anything.
A B2B SaaS auto-tags "Bug Report" on messages matching error code regex like /\bE\d{4}\b/, then routes to the engineering on-call channel in Slack with the customer's plan tier and account ID. Bugs from enterprise accounts get a "P0" sub-label that triggers PagerDuty; everyone else gets a 24-hour SLA.
A telemedicine platform auto-tags "Emergency" when messages match keywords like "chest pain", "bleeding", "unconscious" with high AI-classifier confidence. The label triggers an immediate call to the on-call doctor and a templated message to the patient with the nearest hospital from their location.
A lending app auto-tags messages with attached documents as "KYC-Submitted" and routes to the KYC team. AI classifies whether the document is PAN, Aadhaar or bank statement, then sub-labels accordingly. The dashboard tracks KYC turnaround time and rejection reasons per document type.
An online MBA programme tags students from corporate-sponsored cohorts as "VIP-Corporate" via CRM attribute lookup on inbound. These get a 10-minute SLA, priority assignment to the senior counsellor, and quarterly satisfaction surveys auto-triggered by the label state.
Chat Labels is included on every SabNode workspace. No separate billing, no extra setup — flip it on from your workspace settings.
Plan 10-20 starter labels organised hierarchically. Wachat's setup wizard suggests common patterns by industry (e-commerce, SaaS, healthcare) which most teams adopt with minor tweaks.
Open the rules editor. Write keyword, regex, AI-intent or attribute conditions. Test each rule against the last 30 days of conversations — Wachat shows how many would have matched and which.
Per label, configure overrides: SLA, routing, internal notes, Slack alerts, auto-replies. A label without a behaviour is just a descriptor; a label with behaviours is a control surface.
Run auto-tagging rules across the entire conversation history (or last 90 days) in a background job. This produces meaningful analytics from day one instead of waiting weeks for forward-looking data.
Review the label analytics dashboard. Add new labels for emerging patterns, retire labels with under 10 hits per month, and tune rules with low precision. Treat the taxonomy as a living artefact.
No credit card. No sales call required. Spin up a workspace, plug in a number, and your team is live in under an hour.