What are best practices for WhatsApp customer support?
Operational playbook for WhatsApp support teams: SLAs, routing, tagging, bot handoff, shift handoffs, template discipline, quality monitoring, and metrics — so every conversation drives CSAT, not chaos. Use a unified inbox, set SLAs under 5 minutes, route by skill, tag conversations, preserve bot context on handoff, monitor quality rating weekly, and train agents on mobile-friendly etiquette.
Last updated: 6 July 2026
WhatsApp is where your customers already are — but most support teams treat it like a personal phone with five agents forwarding screenshots. Conversation management on the Business API means shared inbox, SLAs, routing rules, bot handoffs, and quality monitoring so every thread drives CSAT instead of chaos.
This playbook is for support leads and ops managers running WhatsApp at scale — not theoretical CX frameworks. You'll get concrete SLA targets, routing patterns, tag taxonomies, shift handoff templates, and the metrics that predict Meta quality rating drops before your number gets restricted.
Start here if you're in a hurry
- First reply target: under 5 minutes in-hours; bot + auto-reply after hours
- Use a unified inbox on Cloud API — never personal WhatsApp numbers for support
- Preserve bot transcript on handoff — agents never re-ask collected info
- Tag every conversation: intent, product, outcome — monthly reports reveal training gaps
- Inside 24h window: free-form replies. Outside: approved templates only
- Review block rate and report rate weekly — pause bad campaigns before Meta does
Foundation: unified inbox on Cloud API
Personal WhatsApp numbers create silos, no audit trail, and ban risk. Business API with a shared inbox gives you assignment, tags, macros, SLA timers, and multi-agent access on one number.
Benefits deep dive: unified inbox guide · Setup: API India
Hard rule: If an agent replies from their personal phone "to help the customer faster," you lose conversation history, compliance audit, and quality metrics. Route everything through the platform inbox.
1. Define and enforce SLAs
SLAs without visible timers are wishes. Display aging threads in your inbox so agents prioritize correctly.
| Metric | Target (in-hours) | After hours |
|---|---|---|
| First response | < 5 minutes | Bot auto-reply + "We'll respond by 10 AM" |
| Resolution — tracking query | < 1 hour | Bot order lookup if integrated |
| Resolution — refund request | < 24 hours | Ticket created; morning queue |
| Escalation trigger | No reply in 15 min | Supervisor queue alert |
- Configure SLA timers in Ominiflow inbox — red highlight on aging threads
- Auto-escalate VIP tags from CRM to senior agent pool
- Weekly SLA breach report by agent and queue
2. Route conversations intelligently
Wrong routing = customer repeats their story three times. Route by skill, language, campaign source, and CRM data.
- Skill queues: billing, technical, Hindi, enterprise
- VIP tag from CRM → senior agent pool with priority SLA
- Campaign reply routing — know which broadcast triggered the message; agent has campaign context
- Round-robin with assignment lock — two agents never reply simultaneously to same thread
- CTWA ad replies → sales queue with ad creative context
- COD cancel requests → ops queue with order data pre-loaded from Shopify
Automation routing: automation platform
3. Use tags and resolution codes
Standardize tags across the team — freeform tagging becomes useless in reporting within a month.
| Tag type | Examples | Used for |
|---|---|---|
| Intent | pre-sales, support, complaint, feedback | Volume by intent |
| Product | SKU category, service line | Product pain points |
| Outcome | resolved, pending-customer, escalated, refund-issued | Resolution tracking |
| Source | ctwa-ad, cod-campaign, organic-inbound | Channel attribution |
Monthly tag reports reveal top pain points and training gaps. Require outcome tag before closing conversation.
4. Perfect bot-to-agent handoff
Bad handoff is the #1 CSAT killer in bot deployments. The customer already told the bot their order ID — don't make them repeat it.
- Agents read full bot transcript before first reply
- Disclose handoff: "You're now chatting with Priya from support"
- Internal notes visible to agents: "Customer angry about delayed COD — offer ₹50 coupon per policy"
- Handoff button always visible in bot — never trap users in loops
- Fallback to human after 2 failed bot intents
Bot setup: no-code chatbot guide · chatbot for business
5. Template and window discipline
Agents who don't understand the 24-hour window create billing surprises and policy violations.
| Situation | What to send | Category |
|---|---|---|
| Customer messaged within 24h | Free-form text, images, buttons | Session (no template) |
| Customer last messaged 25h ago | Approved template required | Utility or marketing |
| Support follow-up after resolution | Utility template | Utility — not marketing |
| Sale promo to opted-in list | Marketing template | Marketing |
Train agents: never send marketing templates for support follow-ups. Pricing impact: API pricing guide
6. Monitor quality rating weekly
- Block rate and report rate per campaign — pause above 1% block rate
- Template pause alerts in WhatsApp Manager — act before Meta restricts number
- Review complaint conversations — what triggered the report?
- Correlation: high-frequency marketing → quality drop within 2 weeks
Prevention: 5 common API mistakes · ban prevention
7. Shift handoffs and async coverage
Conversations don't end at 6 PM. Async coverage prevents morning backlog disasters.
- End-of-shift summary: open tickets, VIP waits, promised callbacks — hand to next shift lead
- After-hours bot covers FAQ; urgent issues create ticket + morning priority queue
- Macros for common replies — consistent tone, faster replies, fewer typos on mobile
- Snooze with reminder — "Customer asked to call back Thursday 2 PM"
- Don't leave threads unassigned overnight — assign to next shift queue explicitly
8. Train agents on WhatsApp-specific etiquette
Email-trained agents write paragraphs. WhatsApp users skim on mobile.
- Short paragraphs — 2–3 lines max per bubble
- Buttons and quick replies for choices — "Reply 1 for refund, 2 for exchange"
- Voice notes sparingly — prefer text + links; many users in meetings
- Confirm resolution: "Did this solve your issue? Reply YES or NO"
- Never request sensitive data — full card numbers, passwords, OTPs
- Emoji with restraint — one per message max in support context
- Multilingual awareness — Hindi-English mix is normal; match customer language
9. Use AI assist, not AI replacement
Suggested replies and sentiment flags help agents move faster. Keep humans on refunds, complaints, and high-value sales.
| Use AI for | Keep human for |
|---|---|
| Suggested reply drafts | Refund approvals |
| Sentiment detection (angry customer flag) | Complex complaints |
| FAQ lookup from knowledge base | High-value sales closing |
| Conversation summary for handoff | Policy exceptions |
Guide: conversational AI 2026
10. Measure what matters
| Metric | Target benchmark | Action if bad |
|---|---|---|
| First response time | < 5 min (in-hours) | Add agents or improve routing |
| CSAT | 4.2+ / 5 | Review handoff quality and macros |
| Bot containment | 50–70% tier-1 queries | Expand FAQ branches |
| Agent utilization | 12–20 conversations/agent/shift | Balance load across team |
| Reopen rate | < 8% | Improve first-contact resolution |
| Block rate (campaigns) | < 1% | Pause campaign, tighten segment |
Run support on a proper inbox
Explore Ominiflow for Support →
Start Free Trial · unified inbox benefits
FAQ
How many WhatsApp chats can one agent handle? expand_more
Typically 12–20 active conversations per shift with macros and bot deflection. Complex B2B cases may be 5–8.
Should support use personal WhatsApp numbers? expand_more
No. Use the Business API with a shared inbox for audit, assignment, and compliance. Personal numbers create silos and ban risk.
What are best practices for WhatsApp customer support? expand_more
Unified inbox, SLAs under 5 minutes, skill-based routing, standardized tags, bot transcript on handoff, weekly quality monitoring, and mobile-friendly etiquette training.
How do I prevent WhatsApp quality rating drops? expand_more
Only message opted-in contacts, use correct template categories, pause high block-rate campaigns, and escalate complaints quickly before users report.
How fast should businesses reply on WhatsApp? expand_more
First response under 5 minutes during business hours. Use bots and auto-replies after hours with clear expectations on human follow-up timing.
What tags should I use for WhatsApp conversations? expand_more
Standardize intent (pre-sales, support, complaint), product category, and outcome (resolved, escalated, refund-issued). Require outcome tag before closing.
How do I handle bot-to-agent handoff? expand_more
Agent reads full bot transcript, discloses human handoff, uses internal notes for context. Never re-ask information the bot already collected. Last updated July 2026. Conversation management playbook for WhatsApp support teams.
