How to Set Up Automated WhatsApp Customer Support for Your Business
Order confirmations (covered earlier in this series) are one piece. The bigger piece is everything customers ask BEFORE and AFTER ordering — “is this available,” “how much is delivery,” “where’s my order,” “can I exchange this.” This guide covers the full spectrum, from free quick replies to AI agents that handle entire conversations — including ChatPadi, a conversational commerce platform built specifically for this.
3 tiers
Quick replies → rules-based bot → AI conversational agent
24/7
The core value of any tier above “manual” — coverage outside business hours
70-80%
Of WhatsApp enquiries are typically repeat questions automation can handle
Handoff
Every good setup includes an “escape to human” path
This series has covered order confirmations (Template 1-4 from the WhatsApp guide) and the order/payment automation behind them. Customer support is the other half — the conversations that happen before someone orders (“do you have this in blue?”), and the ones that happen after (“I want to return this”). Done well, automation handles the repetitive 70-80% of these — freeing you for the conversations that need a real person.
Tier 1
Free WhatsApp Business app — quick replies, catalogue, away messages
Tier 2
Rules-based bot — keyword triggers, decision trees
Tier 3
AI conversational agent — understands free-form questions, takes orders
All tiers
Need a clear path to a human for anything automation can’t handle
Most businesses move through these as enquiry volume grows
The Core Components of Automated Support
Regardless of which tier you’re at, good WhatsApp support automation covers these — each one addresses a specific category of customer message:
📋 Catalogue / Product Info
Answers “do you have X,” “how much is Y,” “what colours/sizes.” Tier 1’s native WhatsApp Catalogue covers static product listings; Tier 3 AI agents can answer variations and combinations (“blue AND medium AND under GHS 100”).
🚚 Delivery & Payment Info
“How much is delivery to [location],” “do you accept MoMo.” Often the single most repeated question category — a strong candidate for Tier 1 quick replies even before any paid tool.
📦 Order Status Lookup
“Where’s my order?” — connects to the order tracking built in the WooCommerce/Paystack automation guides. Tier 2/3 can look up an order reference and respond with current status automatically.
🔄 Returns & Exchange Policy
“Can I return this?” — a Tier 1 quick reply with your policy is the minimum; Tier 3 agents can walk a customer through initiating a return conversationally.
⏰ Business Hours / Availability
Away messages (Tier 1) set expectations outside hours — “We’ll respond by 9am” reduces frustration even when nobody’s actively replying.
🙋 Escalation to a Human
The component every tier needs: a clear, simple way for the customer (or the bot itself) to flag “this needs a person” — covered in detail below.
DIY (WhatsApp API + Make.com) vs AI Agent Platforms
This series has built several pieces of Tier 2/3 functionality using the WhatsApp Business API + Make.com pattern from earlier guides — order confirmations, status updates. Extending that same DIY approach to full conversational support is possible but involves significantly more complexity: natural language understanding, conversation state (remembering what was discussed earlier in the chat), and catalogue-awareness aren’t things Make.com modules do natively — they’d require connecting to an AI model’s API (Claude, GPT, etc.) with substantial prompt engineering and conversation management logic.
This is the gap that dedicated AI agent platforms for WhatsApp commerce fill — purpose-built for exactly this conversational, catalogue-aware, order-taking use case, without building the conversation-management layer from scratch.
Approach
What it takes
Best for
DIY (Make.com + AI API)
WhatsApp Business API access, Make.com scenario calling an AI model API, custom conversation-state management, ongoing prompt maintenance
Technical teams wanting full control, or very specific custom logic not covered by existing platforms
AI Agent Platform (e.g. ChatPadi)
Sign up, connect WhatsApp number, upload/sync catalogue, configure business details
Most social commerce sellers — purpose-built, guided setup, maintained by the platform
AI conversational commerce for African social media sellers — chatpadi.com
ChatPadi is built specifically for the audience this series has been written for: sellers running their business primarily through WhatsApp and social media, taking payments via Mobile Money and card. Rather than a generic chatbot builder, ChatPadi’s AI agent — called Ama — is designed to handle conversations the way a sharp human shop assistant would: answering product questions from your actual catalogue, discussing price and availability, and guiding a customer to checkout.
Catalogue ManagementSync or manage your product catalogue directly — Ama answers questions based on real, current stock and pricing, not a static script.
Multi-Gateway PaymentsPaystack, Flutterwave, and Stripe (bring-your-own API keys) — generates payment links within the conversation, connecting directly to the gateway setups covered earlier in this series.
Order Lifecycle ManagementOrders taken via WhatsApp conversation are logged and tracked through to fulfilment — the conversational equivalent of the order hub from the automation tools guide.
WhatsApp Channel IntegrationConnects to your business WhatsApp number — customers message the number they already know; Ama handles the conversation.
Pricing tiers (verify current pricing at chatpadi.com):
Starter
Free
Get started
Hustler
$19/mo
Growing sellers
Business
$49/mo
Established stores
Enterprise
$149/mo
Larger operations
How this fits with the rest of the series: If you’ve built the Paystack → Make.com → WhatsApp confirmation pipeline from earlier guides, ChatPadi handles the conversational side that pipeline doesn’t — the questions before checkout. The two aren’t competing; a common setup is ChatPadi (or Ama) handling the conversation and generating the payment link, with the Paystack webhook automation from earlier in this series picking up from there for confirmation, logging, and fulfilment tracking.
Step-by-Step: Setting Up Tiered Support
1
Step 1 · Today, free
Audit Your Last 50 Conversations
Before building anything, scroll through your recent WhatsApp conversations and categorise: how many were repeat questions (price, availability, delivery cost) vs genuinely unique (custom orders, complaints, negotiations)? This number determines how much automation actually helps — and which of the core components above to prioritise.
💡 If repeat questions are under 50%, Tier 1 (free) may be sufficient for now. Above 70%, Tier 3 likely pays for itself quickly in time saved and faster responses.
2
Step 2 · Today, free
Set Up Tier 1 Completely
Regardless of where you end up, Tier 1 is the foundation. In WhatsApp Business: build your Catalogue (every product, price, photo), write Quick Replies for your top 10 repeat questions (delivery costs by area, payment methods, return policy), and set an Away Message with expected response time.
3
Step 3 · Based on Step 1’s findings
Decide: Tier 2, Tier 3, or Stay at Tier 1
If the audit showed mostly simple, predictable questions with low volume — Tier 1 may be enough, revisit in a few months. If volume is high but questions are still mostly predictable — a Tier 2 rules-based bot (via the WhatsApp Business API + Make.com pattern, with keyword-triggered responses) may suffice. If questions are varied, conversational, and involve actual product/order-taking — Tier 3 (an AI agent platform like ChatPadi) is the better fit.
4
Step 4 · If Tier 3
Onboard to an AI Agent Platform
For ChatPadi specifically: sign up at chatpadi.com, connect your WhatsApp Business number, upload or connect your product catalogue, and configure your business details (delivery areas and costs, payment methods, policies) so Ama has accurate information to work from. The Starter tier lets you trial this with your real catalogue before committing to a paid tier.
💡 The accuracy of an AI agent’s answers depends entirely on the accuracy and completeness of the catalogue and business info you provide — spend time getting this right before going live, the same way “garbage in, garbage out” applies to any system.
5
Step 5 · Any tier
Build the Escalation Path
Before going live with any automated tier, define and test how a customer (or the bot) flags “this needs a human” — see the escalation design below. Test it yourself by deliberately asking something the automation can’t handle and confirming the handoff works smoothly.
Designing the Escalation Path
When Automation Should Hand Off to a Human
The difference between helpful automation and a frustrating wall
The specific triggers worth building into any tier:
Explicit request: “agent,” “human,” “speak to someone,” “real person” — should always immediately hand off, no exceptions.
Repeated failure: If the same question is asked 2-3 times without resolution (the customer rephrasing because the first answer didn’t help), hand off rather than continuing to attempt automated answers.
Out-of-scope categories: Complaints, disputes, anything involving “I’m unhappy/angry/this is wrong” — these need empathy and judgment automation shouldn’t attempt.
Low-confidence catalogue/FAQ matches: For AI agents, if the system isn’t confident a catalogue item or FAQ matches the question, handing off (rather than guessing) avoids giving wrong information confidently — which damages trust more than “let me check and get back to you.”
The handoff itself matters: “I don’t understand, please rephrase” repeated multiple times is the #1 complaint about bad chatbots. A graceful handoff — “Let me get one of our team to help with this, they’ll be with you shortly!” — followed by an actual human picking up the conversation, turns a potential frustration point into a non-event. Whatever tier you’re at, make sure a real notification (to you, or your team, e.g. via the Telegram alert pattern from the WooCommerce order management guide) actually fires when a handoff happens — an escalation that nobody sees defeats the purpose.
Which Tier Should You Start With?
🎯 Get a Recommendation
Two questions about your current support load
Recommended starting tier
—
5 Mistakes That Make Automated Support Worse, Not Better
❌ No escalation path at all
A bot that can only say “I don’t understand” with no way to reach a human — especially frustrating when the customer has money ready to spend and a question the bot can’t answer.
→ Fix: Build the escalation path (above) before going live with any tier, and test it yourself by asking something deliberately out-of-scope.
❌ Letting catalogue/policy info go stale
An AI agent confidently telling a customer a product is available when it sold out yesterday, or quoting a delivery price that changed last month, creates more problems than no automation at all — confidently wrong beats “let me check” in trust terms, in the wrong direction.
→ Fix: Whatever platform you use, treat catalogue and policy updates as part of your regular routine — not a one-time setup. For ChatPadi or similar platforms, check how catalogue sync works (real-time vs manual update) and build the update habit around that.
❌ Jumping to Tier 3 without doing the Step 1 audit
Paying for an AI agent platform when your actual enquiry volume is 5/day and mostly unique questions means paying for capability you don’t need yet — Tier 1, free, may genuinely be sufficient.
→ Fix: Do the conversation audit (Step 1) honestly before choosing a tier. Most platforms (including ChatPadi’s Starter tier) have free or low-cost ways to trial before committing to a paid tier — use them to validate fit with real volume.
❌ Automating tone that doesn’t match your brand
A formal, corporate-sounding bot for a brand that’s otherwise warm and personal (or vice versa) creates a jarring experience — customers notice when the “voice” of a business changes depending on whether they’re talking to the bot or a person.
→ Fix: Whichever tier/platform, review and adjust the tone of automated responses to match how you’d actually talk to a customer — most platforms allow configuring this, and it’s worth the time investment.
❌ Treating support automation as separate from order automation
If the AI agent handling support conversations doesn’t connect to the order/payment automation from earlier guides, you end up with two disconnected systems — orders taken via the AI agent that don’t appear in your order log, or order status the support bot can’t actually look up.
→ Fix: When evaluating any Tier 3 platform, specifically check how it connects to your existing order records and payment gateways — for ChatPadi, this is the multi-gateway payment integration and order lifecycle management features that connect directly to the Paystack/Flutterwave setups from earlier in this series.
Frequently Asked Questions
Can I run Tier 1 (free WhatsApp features) alongside an AI agent platform like ChatPadi?
The native WhatsApp Business app’s Catalogue and Quick Replies are features of the app itself — an AI agent platform connects to your WhatsApp Business number via the API, which is a different connection method (covered in the WhatsApp confirmations guide’s discussion of native app vs Business API access). In practice, most businesses moving to an AI agent platform migrate their “voice” — the catalogue info, policies, FAQs — into the agent platform’s configuration, since the agent becomes the primary handler of incoming messages. The native app’s catalogue can still exist for browsing, but the agent typically becomes the conversational layer.
How does an AI agent know what’s actually in stock?
This depends on the platform’s catalogue sync method — some sync with existing e-commerce platforms (WooCommerce, Shopify) for real-time stock levels, others use a manually managed catalogue within the platform itself. For ChatPadi specifically, catalogue management is a core feature — check current documentation for the specific sync methods available, and factor stock-accuracy into the “keeping info current” mistake above regardless of which method you use.
What happens to conversations during the handoff to a human?
Well-designed handoffs preserve conversation history — the human picking up the conversation can see what the customer already discussed with the automated agent, avoiding “as I just told your bot…” frustration. When evaluating any platform, specifically check whether handoff includes conversation context, not just a notification that “someone needs help.”
Is it worth building Tier 2 (DIY rules-based bot) as a middle step, or should I jump from Tier 1 to Tier 3 directly?
Tier 2 makes most sense if your enquiry volume and question variety genuinely sit in the middle — high enough that Tier 1 alone isn’t coping, but predictable enough that a keyword-triggered bot covers most cases without needing AI-level understanding. If your audit (Step 1) shows varied, conversational questions even at moderate volume, Tier 2’s rules-based approach will feel limiting quickly, and jumping to Tier 3 directly avoids building something you’ll likely replace soon after.
Start With the Audit, Not the Tool
Every tier in this guide — including ChatPadi and other AI agent platforms — works best when it’s solving a problem you’ve actually measured (Step 1’s audit), not a problem you’ve assumed exists. Tier 1 is free and takes under an hour; for many small sellers it’s genuinely sufficient. When volume and question variety outgrow it, Tier 3 platforms purpose-built for African social commerce — ChatPadi among them — pick up where DIY automation becomes impractical, and connect directly to the payment and order automation already covered across this series.
Practical guides on automation, online business systems, and growing income from Ghana and across Africa. Visit OurInternetBusiness.com and bookmark it.
