“We were sold a chatbot last year. It’s on the site, nobody uses it, and we don’t even know whether it brings in anything.” We hear that sentence often. The problem is almost never the technology — it’s the choice of channel, the use cases, and the absence of return measurement.
This article doesn’t sell dreams. It explains why a chatbot on WhatsApp or Telegram often outperforms a website widget, which five use cases really make money, what setup actually costs (API + AI + development, concrete numbers), how to measure ROI without lying to yourself, and the three mistakes that turn a promising project into an abandoned gadget.
Target audience: SME executives, e-commerce managers, and marketers considering conversational automation who want an honest decision framework rather than a seductive demo.
Why WhatsApp/Telegram, and not a web widget
The chat widget on a site has a structural flaw: it only works while the visitor is on the page. They leave, the conversation dies. You can’t re-engage them, you don’t have their number, and their attention is captured by ten other tabs.
WhatsApp and Telegram change the equation for three reasons.
First reason — open rates. A WhatsApp message is opened in more than 90% of cases, often within minutes. Marketing email hovers around 20 to 30% open rate. The mobile conversational channel captures attention where email fails.
Second reason — conversation persistence. On WhatsApp, the history stays. The customer can resume a conversation three days later, you can re-engage them (within the rules), and the relationship builds over time. The web widget is memoryless and without follow-up.
Third reason — familiar ground. People already live in WhatsApp and Telegram. You’re not imposing a new interface — you’re arriving in a space where they’re comfortable. Usage friction drops to almost zero.
This doesn’t mean the web widget is useless: it remains relevant for immediate support on a product page. But for lead qualification, follow-up, and long-term relationships, the messaging channel clearly outperforms.
5 use cases that make money
Here are the five cases where a messaging chatbot generates measurable return — not gadgetry.
1. Lead qualification 24/7
A prospect arrives outside business hours. The bot engages, asks 3 to 5 key questions (need, budget, timeline), qualifies, and hands the salesperson a hot lead with its context the next morning. You no longer lose the 10pm leads. For a service business, capturing even 20% more leads at night and on weekends changes the commercial equation.
2. Level 1 support
“Where’s my order?”, “What are your hours?”, “How do I return a product?” — 60 to 80% of incoming requests are repetitive. A well-built bot handles them instantly, at any hour, and only escalates complex cases to a human. The result: your support team focuses on what has value, and the customer response time goes from hours to seconds.
3. Abandoned cart recovery
E-commerce loses about 70% of carts along the way. A WhatsApp message sent an hour after abandonment — “There’s still an item in your cart, would you like to finish?” — recovers a significant percentage of these sales, with an open rate incomparable to email. It’s one of the most direct ROIs in all of conversational e-commerce.
4. Appointment booking
For appointment-based activities (practices, services, consulting), the bot proposes slots, confirms, sends a reminder the day before, and handles rescheduling. It makes phone tag disappear and drastically reduces no-shows thanks to automatic reminders.
5. Upsell and post-purchase follow-up
After a sale, the bot sends delivery tracking, requests a review, and proposes a complementary product at the right moment. Post-purchase is when the customer is most engaged — and most neglected by most companies. A well-designed post-purchase flow increases customer lifetime value with no additional sales effort.
Tech stack: what’s under the hood
A professional messaging chatbot rests on three building blocks.
The channel layer. For WhatsApp, it’s the official WhatsApp Business API (via a provider like Meta directly, Twilio, or 360dialog). Careful: the free “WhatsApp Business” of the mobile app isn’t enough for automation at scale — you need the official API, which involves account validation and per-conversation pricing. For Telegram, the Bot API is free and far simpler to implement.
The intelligence layer. This is where the LLM (language model — GPT, Claude, or equivalent) comes in. It understands the request in natural language, generates the answer, and knows when it doesn’t know (and must escalate to a human). The quality of the “system prompt” and the knowledge base determines 80% of perceived quality.
The integration layer. The bot doesn’t live alone: it connects to the CRM (to create/update a lead), to the store (to know an order’s status), to the calendar (for appointments). Without these integrations, the bot stays a demo. With them, it becomes an employee.
Real costs: the honest estimate
Let’s break down the real cost of a WhatsApp chatbot for an e-commerce SME.
Setup cost (one-off): conversational design, development, CRM/store integrations, knowledge base, testing. Depending on complexity, count €4,000 to €15,000. A simple Telegram bot costs less; a multi-integration WhatsApp bot costs more.
WhatsApp conversation cost (recurring): Meta charges per conversation (24h window), with rates varying by country and type (marketing, utility, service). In Europe, count a few cents per utility conversation. At 2,000 conversations/month, we’re talking tens of euros, not thousands. Telegram is free on the channel side.
AI cost (recurring): each exchange consumes LLM tokens. With a modern model, the cost per conversation is fractions of a cent to a few cents depending on length. At an SME scale, it’s generally €10 to €100/month.
Maintenance (recurring): knowledge base updates, adjustments, supervision. Count €200 to €600/month depending on volume and ambition.
In summary: a serious project starts around €4,000 to €8,000 in setup, then a few hundred euros per month in run. If you recover even 5 abandoned carts or 3 more qualified leads per month, the return is generally reached within a few months.
Measuring ROI without lying to yourself
A chatbot you don’t measure is a cost, not an investment. Here are the indicators that really matter.
- Deflection rate: percentage of requests resolved by the bot without human intervention. This is the direct saving on support.
- Qualified leads generated: how many leads the bot captured and transmitted, and how many converted. To compare against total cost.
- Carts recovered: value of orders finalised following a bot re-engagement. This is money that would have been lost.
- Time to first response: before/after. Going from hours to seconds impacts conversion and satisfaction.
- Escalation rate and satisfaction: if the bot escalates too much, it’s poorly calibrated; if it escalates too little, it frustrates. The balance is measurable.
The classic mistake: judging a bot by “number of messages handled.” What matters is money earned and saved, not volume.
3 mistakes to avoid
Mistake 1 — trying to automate everything from the start. A bot that tries to do everything fails everywhere. Start with a high-ROI use case (abandoned cart, or lead qualification), master it, then expand. Scope is won through iterations, not big bang.
Mistake 2 — hiding that it’s a bot, or blocking access to a human. Customers tolerate a bot if it’s useful and if they can reach a human when they need to. A bot that traps the user in a loop with no exit destroys the relationship. The “talk to an advisor” button must always exist.
Mistake 3 — ignoring GDPR. You process personal data in a conversation. Consent, purpose, retention, right to erasure: all of this is designed from the start, not as an afterthought. A non-compliant chatbot is a legal risk, not an asset.
In practice
A messaging chatbot is neither a magic wand nor a gadget. It’s a digital employee that, well-targeted on the right use cases and properly integrated with your tools, makes money measurably. The difference between a project that succeeds and one that ends up abandoned comes down to three things: the right channel, the right use cases, and return measurement.
At Seganiko, we design and develop chatbots (WhatsApp, Telegram) and AI assistants integrated with your CRM and your store, in France and Luxembourg. We always start by identifying the highest-ROI use case for your business — and we estimate the return before developing.
Evaluate the ROI of a chatbot for your business
Need to improve your SEO?
We audit your site and implement a measurable SEO strategy.
See our SEO service →
