Your next storefront is a chat window
Customers already live in WhatsApp. The brands winning there didn’t build a chatbot. They put their workflows behind a conversation.
- Start with one workflow, such as ordering, warranty claims or bookings, not a chatbot personality.
- Understand what people mean first; use menus and buttons only as shortcuts.
- Connect to the systems that do the work, such as the CRM, store or ticketing tool.
- Measure problems resolved, not messages sent.
Picture a customer with a speaker that has stopped working. The old journey: find the support page, fill in a form, receive a ticket number, wait for an email, reply to the email, wait again. Somewhere around step four they write a review instead.
The new journey: they send a message, in the app they already have open, to tell you what’s wrong. Then the business does the rest.
That second journey is what we build with Talking Shops, our platform that turns business workflows into conversations. AI works out what the customer wants and then actually does it. It’s now live with more than 30 brands across retail, consumer electronics, legal services, food and beverage, and health. Here’s what those deployments taught us.
What it looks like in practice
Marshall sends customers who report a product issue to a WhatsApp conversation powered by Talking Shops. It understands the problem, creates a ticket in Zoho and orchestrates the resolution from start to finish. The customer never sees a form.
Da Milano runs customer support and shopping through chat on both its website and WhatsApp, so the same conversation can answer a question and close a sale.
Brands like Just One and Sage Organics use chat-based shopping. Elnagger & Partners uses it for legal and corporate advisory enquiries, Varak for food and beverage, and others for health and nutrition consultations. Different sectors, same idea: let the customer say what they need in their own words.
A bot that can’t take action is just an FAQ with a typing indicator.
Six lessons from 30+ deployments
1. Start with a workflow, not a personality
The first question isn’t “what should our bot be called?” It’s “which process do customers hate most?” Warranty claims, order tracking, bookings and reorders are good places to start. Give it a personality later, if at all.
2. Understand the request first, menus last
“Press 1 for orders” rebuilt the phone tree inside a chat app. Let people type what they mean, understand it, and offer buttons as shortcuts rather than the only way through.
3. Connect to the system of record
The value comes from what happens after the message. Creating the ticket, checking stock, updating the booking. That means real connections to the CRM, the store and the helpdesk, not a screen-scraper and a prayer.
4. Design the handoff to a human
Some conversations need a person. Make the switch seamless, with the full context passed along, so the customer never has to say “as I already explained” twice.
5. Respect the channel
WhatsApp is where people talk to their families. Ask permission before you message, keep templates useful, and never treat it like a bulk email list. The channel’s rules are strict for good reason.
6. Measure problems resolved, not messages sent
More messages can mean a confused customer. Track issues resolved without a human, time to resolution, conversion from chat and repeat contact. Those are the numbers that show up in revenue and in reviews.
Pick the one workflow with the highest volume and the worst current experience. Put it in a conversation, connect it to the system that does the work, and measure resolution for 30 days. Then do the next one.
The storefront hasn’t disappeared. It’s just moved into the place your customers already spend their day, and it would like a reply.
