Conversational AI in retail pays off when your staff spend their day answering the same five questions: is it in stock, when do you open, can I book a fitting, where is my order, can I return this. This article covers independent retailers, showroom businesses and multi-site shops with 10 to 250 staff. You will set up an AI agent that answers those questions on chat, SMS and social messages, books in-store appointments, sends payment links and asks happy customers for a review.
The retail problem it solves
A shop's busiest selling hours are also its busiest messaging hours. The person who should be helping a customer at the counter is answering an Instagram message about Saturday opening times. Messages that arrive after closing wait until the next morning, and some buyers simply go elsewhere.
An agent answers the routine questions instantly and around the clock. Your floor staff deal with the customers in front of them. The conversations that need a person, such as a damaged delivery or a large trade order, reach the right person with the history attached.
Before you start
- An account with your social accounts, web chat and SMS number connected to Conversations.
- Your opening hours per location, delivery and returns policy, and the questions customers ask most.
- A calendar for each bookable service, such as fittings, consultations or showroom visits.
- Products or services set up in Payments if you want the agent to send payment links.
- A named person per location who receives hand-offs.
How to set it up
- Go to AI Agents > Conversation AI and open Agents List.
- Select the bot, then click Deploy.
- Turn on the channel you need.
- Open the channel's edit icon if it has extra settings.
- Turn on Assign channel and choose the page, number, account or widget.
- Add any Has tags and Doesn't have tags conditions.
- Click Update, then click Save for the bot.
- Repeat for each additional channel and test the result.
Stock questions need care
Stock is where retail agents go wrong. Unless the agent can see live stock, it should never say an item is available. Give it a safe answer instead: it takes the product name and size, records them on the contact and promises a reply from the shop within a set time.
A workflow then creates a task for the location. A staff member checks the shelf and replies from the same conversation. The customer gets one thread, and the shop gets a list of items people asked for, which is useful buying data.
Imagine a retailer receiving 900 messages a month across Instagram, Facebook and web chat. If about two thirds are about opening hours, delivery areas and showroom booking, the agent could answer those 600 and book the visits. The remaining 300 or so, mostly stock and custom orders, would reach the location team with the details already captured.
What it replaces and what it costs your team
It replaces the phone ringing for opening hours, the shared social inbox nobody owns and the paper diary for appointments. Expect to spend around an hour a week reading hand-offs and adding answers. In peak season, check daily, because new questions appear with each promotion.
Usage such as SMS and AI replies is billed as used on the platform, so a quiet month costs less than a busy one. There is no per-seat charge for adding shop staff to the inbox.
Retailers often launch the agent the week a sale starts. That is the worst week to learn. Launch it in a quiet period, fix the gaps, then let it handle the peak.
When not to use it
If most of your sales are high-value and consultative, such as bespoke jewellery, customers may prefer to talk to a named person from the first message. Use the agent only for booking in that case. If you sell mostly online through a separate store platform with its own order tracking, keep order status questions there and use the agent for showroom and service conversations.
For a broader view of how retailers put chat on their site, see chat bots for websites. For text-message replies specifically, read AI text messaging.
Troubleshooting
The agent gives one shop's hours for another. Each location needs its own clearly labelled entry in the knowledge base. Ask the agent to confirm which shop the customer means.
Customers book appointments that clash with staff rotas. Calendar availability is wrong, not the agent. Set availability per staff member and block holidays in the calendar.
Refund requests get automated replies. Add refunds and returns disputes to the hand-off list by name.
Review requests go to unhappy customers. Add a condition so the workflow skips contacts who had a complaint tagged in the last month.
Frequently asked questions
Can conversational AI in retail take orders?
It can send a payment link for products or deposits you have set up in Payments. Full basket checkout belongs on your online store. Use the agent to answer questions and point buyers to the right place.
Does it work for multiple shops?
Yes. One agent can cover several locations if each has its own hours, calendar and hand-off person. Customers tell it which shop, or it asks.
Will it reply on Instagram and Facebook?
Yes, once those accounts are connected to Conversations. All replies sit in one inbox. More channel detail is in omnichannel conversational AI.
What should we measure?
Count appointments booked through chat, messages answered without a hand-off and reviews collected. Those three tell you whether staff time has actually moved back to the shop floor. Examples of good exchanges are in conversational chatbot examples.