CASE STUDY · FIELD SERVICES AI (SMB)
An AI booking & lead-qualification agent for a field-service SMB network.
- 18%Missed calls down from 18% to 9%
- 12%Booked jobs up 12%
01 · PROJECT OVERVIEW
The client and the brief.
An anonymised network of 3 field-service branches. Name withheld under confidentiality.
A voice and chat agent that qualifies leads, books into the calendar and escalates urgent jobs. It answers every inbound call and web message, asks the questions the office would ask, offers available slots and confirms the booking by text.
- 118% of inbound calls went unanswered after hours.
- 2Each branch handled its own phone line with a small office team during the day.
- 3Evenings and weekends rolled to voicemail, and few customers left messages.
- 4The owners knew they were losing work but had no way to see how much, and no appetite for a night-shift call centre.
02 · WHAT WENT WRONG
What went wrong, and how we fixed it.
Every engagement hits problems. These were the ones that mattered here, and what we changed so they did not come back.
- PROBLEM 01In the first week of the supervised run, the agent misjudged urgency on calls where customers described problems in indirect ways, such as "the water is a bit brown" or "it smells odd"
Some of those were routine and some were not. We rewrote the urgency questions to ask directly about specific signs, added a rule that any mention of gas, flooding or no heat goes to a person, and tested the new flow with recorded examples. We also made the agent err towards hand-off in doubtful cases.
- PROBLEM 02A few extra transfers were an acceptable price
03 · THE PROCESS
How we ran it.
- 01Discovery
We listened to recorded calls from each branch and wrote down the usual questions, the job types, the service areas and what counted as urgent. The branches differed more than the owners expected, so we kept shared rules and local settings apart.
- 02Build
Call flows and calendar integration in short iterations, with test calls from branch staff acting as awkward customers.
- 03Supervised run
One branch went first. Staff reviewed transcripts every day and marked errors. Corrections went into the script and the rules.
- 04Rollout
The other two branches followed, with weekly reviews for the first weeks and then a monthly check.
04 · RESULTS
What changed.
Measured against the way the work ran before the engagement began.
- NOW 01
Branch managers read the overnight summary each morning instead of checking voicemail.
- NOW 02
Staff spend less time on routine booking calls and more on quotes and customer visits.
- NOW 03
The owners review weekly figures for missed calls and bookings by branch.
- NOW 04
Script changes follow a simple review: the manager proposes, the team tests, then it goes live.
- 18%
Missed calls down from 18% to 9%
- 12%
Booked jobs up 12%
05 · WHAT WE BUILT
How it works.
- Telephony routes inbound calls to the agent when the office is closed, busy or unattended.Daytime routing is configurable per branch.
- A speech layer transcribes the caller, a language model drives the conversation within a script and rule set, and a voice layer replies.
- The agent follows a structured flow: identify the need, check the service area, assess urgency, collect contact details, offer slots.
- Booking writes to each branch's calendar through its API, with checks to prevent double-booking.
- Urgent jobs trigger a live transfer or an alert to the on-call technician.
- Every call produces a transcript, a summary and a structured lead record.Call recordings follow the client's retention rules.
- Controls and handover
- Branch managers could pause the agent for their own line with a single switch, for example during a local outage or an unusual event. Call recordings were kept only for the period the client set, and listening access was limited to managers. At handover we gave each branch a one-page guide to reading the morning summary and a short list of changes managers may make themselves, such as opening hours and service areas.
06 · QUICK ANSWERS
The short version.
- What problem did the client have?
- 18% of inbound calls went unanswered after hours. Each branch handled its own phone line with a small office team during the day. Evenings and weekends rolled to voicemail, and few customers left messages. The owners knew they were losing work but had no way to see how much, and no appetite for a night-shift call centre.
- What did Brandsmashers build?
- A voice and chat agent that qualifies leads, books into the calendar and escalates urgent jobs. It answers every inbound call and web message, asks the questions the office would ask, offers available slots and confirms the booking by text.
- What were the results?
- Missed calls down from 18% to 9%; booked jobs up 12%.
YOUR TURN
Have a similar requirement?
Tell us what is stuck. We will come back with a plan and matched, pre-vetted engineers.
