How to Build a WhatsApp Chatbot with Respond.io
You can build a working WhatsApp lead-qualification chatbot in Respond.io in about two hours: create a flow triggered by inbound message, add button-based questions, score the answers, and route qualified contacts to a human with an SLA clock. The build is fast; the Meta template approval in the middle is what determines your timeline.
The most common failure is not technical. It is building a bot that talks too much. WhatsApp is a personal channel and users abandon chatty automation within two messages.
What should the bot collect?
Three fields, maximum. Any more and completion drops sharply.
| Field | How to ask | Why it matters |
|---|---|---|
| Intent | Buttons: Pricing / Support / Partnership | Routing |
| Company size | Buttons: 1-10 / 11-50 / 51+ | Lead scoring |
| Timing | Buttons: Now / This quarter / Later | Priority |
Notice every question is a button. Free-text fields on WhatsApp get answered with emoji, voice notes, and half-sentences. Buttons produce structured data you can act on.
How do you build the flow in Respond.io?
Work through the four stages below in one sitting.
- Create the trigger. New message received → channel is your WhatsApp number. Do not trigger on every message; add a condition that the contact has no prior conversation, or returning customers get re-interviewed.
- Add the qualification nodes. One message per question, buttons only, with a short progress hint on the second question (“Last one”). Users finish more often when they can see the end.
- Add a score step. Respond.io lets you set contact fields from flow answers. Map answers to a numeric score: Pricing + 51+ employees + Now equals a hot lead.
- Add the routing branch. Score above threshold → assign to sales with a 30-minute SLA. Below threshold → assign to a nurture queue with a 24-hour SLA and a softer intro message.
Test the whole flow on an internal number before pointing it at customers. Button payloads fail silently more often than you would expect.
How do the Meta message templates work?
This is the part that stops projects. WhatsApp only allows free-form messages inside a 24-hour window that starts when the customer messages you. Outside that window you need a pre-approved template.
Practical rules:
- Submit templates early. Approval usually takes minutes but can take 24 hours, and rejections are common.
- Avoid promotional language. Words like “offer”, “discount”, and excessive capitalization trigger review or rejection under Meta’s marketing template policy.
- Keep variables minimal. Templates with more than two variables get rejected more often.
- Categorize correctly. A utility template sent as marketing gets flagged. Follow-up messages about an existing inquiry are utility.
Your bot’s follow-up at hour 25 needs a template. Plan for it or your follow-up silently fails.
How do you keep the bot from feeling robotic?
Write messages like a person texting, not like a form. “Quick one before I pass you to a human — what’s this about?” beats “Please select your inquiry category.”
Add one escape hatch in the first message: “type agent anytime to talk to a person.” Then honor it. Nothing kills trust faster than a bot that ignores the escape word.
Set the bot to only run during business hours if you cannot staff the handoff. A bot that qualifies a hot lead at 2am and routes it to an empty queue is worse than no bot.
What results should you expect?
Teams that keep qualification to three button questions typically see 55-70% completion. Add a fourth free-text question and completion drops into the 30s. Those figures come from Respond.io users sharing flow analytics, and they track with general chatbot abandonment research.
On routing, the win is response time. A bot that assigns a qualified lead within 30 seconds cuts first-response time from hours to minutes, and in WhatsApp sales, first response correlates strongly with conversion.
When is a WhatsApp bot the wrong choice?
If your average deal is above roughly $50,000, the qualification conversation needs a human from message one. Enterprise buyers read automated WhatsApp as a signal that you are small or understaffed.
Same for regulated industries — financial advice, medical questions, legal. The compliance review for a bot that improvises wording will cost more than the bot saves.
Key Takeaways
- A Respond.io WhatsApp qualification bot takes about two hours to build and four button questions maximum.
- Keep qualification to three button-based fields; free text cuts completion roughly in half.
- Meta template approval gates follow-ups outside the 24-hour window — submit early and avoid promotional wording.
- Route on a numeric score with a 30-minute SLA for hot leads and a 24-hour queue for the rest.
- Skip the bot entirely for $50k+ enterprise deals and regulated topics.
Start with the Respond.io overview at iAIFeed if you have not connected a channel yet.
