WhatsApp is the most important sales channel in Costa Rica. If your business has an active WhatsApp line, more demand lands there than anywhere else. The question isn't whether to automate — it's where. These are the cases where the bot wins, the cases where it loses, and how to decide.
💰 How much does a WhatsApp bot cost in CR? See the complete pricing guide — conversational bots USD 1,500–3,000, AI pipelines USD 3,000–5,000, agents with tools USD 5,000–8,000.
Where the bot beats the human (with data)
1. 24/7 initial qualification
35–40% of leads landing on WhatsApp arrive outside business hours: night, Saturday afternoon, Sunday. If the team replies Monday at 9 am, 36 hours have already passed — and in that time the lead tried the competitor that responded at 10 pm.
A qualifier bot solves this. It asks the basics (name, what you want, rough budget, when you need it), saves the conversation, and tells the customer a human confirms Monday with a proposal. The lead stays "warm" instead of "cold", and the team wakes up Monday with 8 qualified leads instead of 8 messages without context.
A client of ours saw lead-to-sale conversion rise 22% just from implementing 24/7 qualification. Bot cost: USD 2,100. Recovered in 3 sales.
2. FAQs with known answers
If 60%+ of incoming questions are variations of the same 10 questions ("where are you?", "what hours?", "do you ship?", "price of service X?"), the bot answers them in seconds. The human is freed for conversations where they actually add value.
A bakery we worked with receives 200 messages a day. 75% are about hours and product availability. Before, two full-time people responded. With the bot, one person supervises and only handles unresolved cases. Savings: USD 600/month in staff cost, bot recovered in 6 months.
3. Scheduling and routing
"I want to book a consult Tuesday afternoon" is a conversation the bot can resolve without a human if it has calendar access. Checks availability, proposes slots, confirms, sends a reminder the day before. The human only steps in if there's a conflict or change.
For a veterinary clinic, the bot books 80% of appointments. The receptionist is freed to handle walk-ins and emergencies.
Where the bot loses (and kills the sale)
1. Angry customers or real pain
If someone writes "my order is 4 days late and I need this NOW", don't treat them with a bot saying "I'm sorry for the delay, can you wait 24 hours?". That enrages. Angry customer = immediate routing to a person, no bot in between.
The technical signal: simple sentiment analysis. Words like "angry", "horrible", "cancel", "doesn't work", "days late" trigger automatic escalation.
2. Big or irreversible decisions
USD 1,000+ purchase. Contract signing. Service cancellation. These decisions almost always lead to a human conversation before closing. If the bot tries to close them alone, it looks robotic and cold, and often the customer walks.
Rule: the bot prepares the big decision (gathers info, books a call), but doesn't close it.
3. Unique or creative conversations
If your business depends on unique-per-client proposals (consulting, design, corporate events), the bot won't improvise well. Better: bot captures brief and books the meeting, human leads the proposal conversation.
The winning pattern: bot + human hybrid
In the last 14 WhatsApp projects we've done, the bot NEVER is the only channel. It's always bot for repetitive cases + human for cases where the human adds value. The most-used routing rule:
Message arrives
↓
[Intent analysis]
├─ Known FAQ → bot answers
├─ Qualify lead → bot asks + saves → notifies human
├─ Schedule → bot resolves if slot exists
├─ Angry customer → immediate human route
├─ Off-catalog case → human route with summary
└─ Purchase > threshold → human route with qualification done
This resolves 70–80% without human intervention, and only hands the team the cases worth the 5–10 minutes of attention.
Cost and time
| Type | Cost (USD) | Time | Recommended messages/month |
|---|---|---|---|
| Basic bot (FAQs + routing) | 1,500 – 3,000 | 3–4 wk | 200 – 2,000 |
| Qualification + scheduling + CRM | 3,000 – 5,000 | 4–6 wk | 1,000 – 10,000 |
| Agent with tools (quotes, invoices) | 5,000 – 8,000 | 6–10 wk | 5,000+ |
Add USD 5–80/month for LLM tokens depending on volume. And USD 30–100/month for WhatsApp Business API if you have high volume of initiated conversations.
The right question before investing
It's not "do I want a bot?". It's "what percentage of my messages are repetitive and would be resolved in 30 seconds by a person if they had context?". If that number tops 50%, the bot pays. If it's under 30%, not yet — invest in a better human process first.
If you're thinking about automating WhatsApp and want to know if it makes sense, let's talk. We'll tell you honestly if it's worth it now or if it's still too early.
