The customer-facing side of Norman's Electronics
Web Service Pay
~8 calls/day
handled first by the advisor
Nightly
cost averages from real jobs
Separated
public site and customer data
The challenge
About eight times a day someone called to ask whether a repair was even worth it. Each call ran twenty minutes, and ten voicemails stacked up meanwhile.
Collecting deposits and card details over the phone was slow and risky.
The approach
- Kept the public site apart from the CRM so customer data stays protected, then passed clean requests across.
- Put a decision tree and real repair history underneath the AI so it gives consistent, explainable answers.
What was built
Service request intake
Customers submit requests with SMS opt-in. The CRM creates or matches the customer and adds the ticket.
Secure card capture
Cards are tokenized for deposits and later charges, with refunds handled in the same flow.
Service catalog and status lookup
Brands and models the shop repairs, with average costs drawn from past jobs and refreshed nightly.
Repair or Replace advisor
AI turns 'it pops and smells like burnt metal' into structured answers, checks repair history and comparable used prices, and quotes shipping both ways.
What's your bottleneck?
Thirty minutes, no pitch deck. Bring the thing that eats your week.