The verdict in three sentences
A WhatsApp chatbot isn't there to replace your sales staff but to absorb the repetitive: prices, stock, hours, order status. Designed well, it handles up to 70 % of messages with no human input and drops agent time by 40 to 60 %. The secret of a good bot isn't AI alone but the handoff rule: knowing exactly when to pass to a human before frustrating the customer.
What automates, what stays human
Not all flows are equal. Some benefit from full automation, others need a salesperson's judgment. Here's the split that works.
| Flow | Automated (bot) | Human (agent) |
|---|---|---|
| FAQ price / hours / delivery | Yes, instant | — |
| Catalog browsing | Yes, guided | — |
| Simple order-taking | Yes, 24/7 | Complex cases |
| Stock check | Yes, if connected | Out-of-stock / pre-order |
| Pay-link | Yes, auto-generated | — |
| Order status | Yes, auto query | Dispute, delay |
| Negotiation / advice | No | Yes, added value |
| Sensitive complaint / support | No | Yes, escalation |
The right architecture lets the bot handle volume and frees humans for value-added selling and tricky cases. You never hide that a human is available: a simple \"talk to an agent\" must always be reachable.
Cost and gains: the 2026 numbers
A chatbot means a setup investment then a modest running cost. Here are orders of magnitude for an African store.
| Item | 2026 order of magnitude | Detail |
|---|---|---|
| Simple chatbot setup | ₦200,000 - ₦500,000 | FAQ + order-taking |
| Advanced AI chatbot setup | ₦500,000 - ₦1,200,000 | NLU + stock integration |
| Monthly maintenance | ₦30,000 - ₦90,000 | Scenario updates |
| API message cost | ₦40,000 - ₦90,000/month | By volume |
| Agent time saved | 40 to 60 % | On repetitive tasks |
| Availability | 24/7 | vs opening hours |
The return comes mostly from orders captured at night and on weekends, when no agent is around, and from time given back to the team to sell instead of repeating the same answers.
Mini case study
Ama runs a deli in Accra receiving 900 WhatsApp messages a month, 60 % of them repetitive (price, availability, delivery hours). Before the bot, one employee spent 4 hours a day on it. After deploying a chatbot at GHS 5,000 setup, the bot handles 63 % of messages alone and takes simple orders at night. Agent time drops from 4 h to 1 h 40 a day, about 50 hours/month freed. At GHS 25/hour, that's GHS 1,250 of value recovered monthly, plus around twenty night orders captured at a GHS 300 average basket. The setup pays back in a little over two months.
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
FAQ
What share of messages can a chatbot handle alone?
A well-tuned bot absorbs 50 to 70 % of messages, mainly FAQs and simple orders. The rest hands off to a human under precise rules.
Does the chatbot work without the paid API?
Advanced automation needs the Cloud API. The free App only offers basic auto-replies, not enough for real order-taking.
When should the bot hand off to a human?
As soon as it detects a complaint, a negotiation, an out-of-stock or three misunderstandings in a row. A \"talk to an agent\" button stays visible at all times.
How long to deploy a chatbot?
A simple FAQ + order bot ships in 1 to 2 weeks. An AI version connected to stock needs 3 to 4 weeks of development and testing.
Does the customer know they're talking to a bot?
Yes, transparency is recommended and often required. A clear tone and permanent human access avoid frustration and protect your reputation.
Let's talk about your project. We design a chatbot that takes your orders 24/7 and knows when to call a human. WhatsApp +221 77 596 93 33.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.
