Customer project

WhatsApp AI customer service for an iPhone retailer

Client

iPhone Retailer (New & Second)

Project Type

WhatsApp AI CS & sales assist (iPhone retail)

Before

The shop was flooded with the same WhatsApp questions all day: stock, price, instalments, warranty, location, trade-in. Staff were serving customers on the floor, chats piled up, and buyers moved on to the shop next door. Around 230 messages arrived per day, and 30% of new leads showed up outside working hours (21:00-09:00).

After

Chats now get answered in under a minute, around the clock, always from real stock data. In the first 80 days the assistant answered roughly 18,000 customer messages - 99% of all outbound replies - and the chat channel accounted for about 1 in 9 of the shop's buyers. Staff still take over for negotiation, complaints, and closing in store.

Bridge

We wired the WhatsApp agent into the shop's inventory database, wrote down explicitly what it may and may not say (negotiation happens in store only, never quote stock levels, never promise a final instalment figure), added tools for credit simulation, location and trade-in estimates, and set up automatic escalation to staff the moment a buyer turns serious or arrives at the mall.

Challenges

Repetitive chats pile up at peak hours

Stock, price and instalment questions arrive non-stop while staff are serving the customer standing in front of them.

Answers must follow stock data, not staff memory

Prices and availability change daily. A wrong answer only surfaces once the customer has already travelled to the shop.

A third of new leads arrive after hours

30% of new leads land between 21:00 and 09:00, when nobody is there to reply.

Solution Highlights

Every answer read from the inventory database

Availability, price, colour, storage, warranty and battery condition are all checked against the shop's database before anything is sent to the customer.

A sales flow, not just an FAQ

Zero-down credit simulation, trade-in and buyback estimates given as ranges, then an invitation to visit. The moment a customer names a day, the agent closes with an appointment template.

Clear limits and human escalation

Negotiation and payment stay in store. Serious buyers, complaints, and customers who have already arrived are routed to staff automatically.

Client Introduction

An iPhone retailer selling both new and second-hand units, with heavy WhatsApp traffic. Almost every transaction still closes in the shop - chat is the front door, and that door was least attended exactly when the shop was busiest.

Project Journey

We started from how the team answered on day one: which messages were pure stock checks, which genuinely needed a human, and which rules had only ever lived in the owner's head.

Once those rules were written down explicitly, the agent was wired to live stock data and rolled out in stages. Two rounds of owner feedback in the first fortnight changed the answering style considerably: the share of chatters who went on to buy rose roughly threefold after that tuning, and stayed there.

Core Development

What we shipped:

A WhatsApp agent that reads stock straight from the shop database before stating any price, colour, storage, warranty or unit condition
Zero-down 12-month credit simulations delivered as an image, with no instalment maths improvised in chat
Trade-in and buyback estimates expressed as price ranges, following the shop's own formulas
Location and opening-hours tools, plus escalation to staff for serious buyers, complaints, and customers already on site
An operator dashboard so staff can take over any conversation at any time

Proof Note

Measured by matching customer WhatsApp numbers against the shop's POS transaction data over the first 2.5 months - not inferred from chat sentiment. The result: about 1 in 9 of the shop's buyers had passed through chat, and the chat channel's share of monthly revenue rose from 5% to roughly 10% and held there. The rupiah figures belong to the client; only the ratios are published here.

Proof Snapshot

~1 in 9 store buyers arrived through chat

Measured by matching customer numbers against the shop's transaction data: in the first 2.5 months, about 1 in 9 of the shop's buyers had come through chat, and chat's share of monthly revenue rose from 5% to roughly 10% and held there.

iPhone Retailer (New & Second) — Measured result, first 2.5 months

Want a similar result in your business?

Book the automation audit and we will map the best workflow to fix first.

Book Audit