Logo
Case study
Trades · Commercial electrical

Over four times the leads, in fewer owner hours

How a 12-person electrical contractor swapped Friday-afternoon prospecting for an AI sales team in Claude Code, securing 12 client meetings in 90 days.

240Companies scored in 90 days, up from 52
68 hrsOwner time, down from 78
12Client meetings secured, up from 1

01 At a glance

The short version

A 12-person commercial electrical firm lived on referrals and one hunt for new work each Friday. We set up an AI sales team in Claude Code that researches each company, scores it and drafts the outreach, while the owner stays in charge of every send.

The client

Commercial electrical contractor

Owner-led, 12 staff. Fits out and maintains offices, shops and warehouses. Sells to property managers, builders and facilities managers.

The pain

Prospecting waits for quiet days

Each lead took about 90 minutes to research. One email went out, then nothing. Nobody could tell which leads were worth a quote.

The solution

Five agents on every lead

One command sends five agents to research a company and score it out of 100. Further commands draft the emails, the follow-ups and the meeting brief.

The outcome

More leads, fewer hours

240 companies scored in 90 days instead of 52, in 68 owner hours instead of 78, securing 12 meetings.

The shift

Before: the owner researches one lead at a time, sends one email and moves on. After: five agents research and score every lead, and the owner spends the hours on the half that are worth a call.

02 The pain

The work is good.
The pipeline is luck.

The firm wins jobs when it gets in front of the right people. It rarely has the time to get in front of new ones.

01

Prospecting waits for a quiet Friday

New-business time only exists when site work eases off. In a busy month it never happens, and most jobs still arrive by referral.

02

Each lead costs about 90 minutes

Reading a company's site, finding the right person and working out what they need takes an hour and a half. Six hours a week buys four companies.

03

One email, then silence

Most leads get a single message. A second or third goes out only if the owner remembers, so good prospects go cold.

04

No way to rank the leads

On paper, every lead looks the same. The owner quotes on instinct, and some quotes eat half a day.

The real cost. A half-day quote for a job that never goes ahead is half a day lost from site work and from the next lead. With no score on each prospect, there is no way to spot those jobs before the quote.

03 The solution

Five agents on every lead,
one owner in charge

Inside Claude Code, type /sales prospect and a company's web address. Five agents research it at the same time and send back a scored report and an outreach plan.

Step 1 · Define the target

/sales icp writes a profile of the ideal customer. For this firm: property managers, builders and facilities managers with commercial sites nearby.

Step 2 · Score each company

/sales prospect sends five agents to look at the company, its people, the job, the rivals and the angle. One score out of 100 comes back.

Step 3 · Reach out

outreach, followup, prep and proposal turn a good score into emails, meeting notes and a draft proposal.

Company web addressone link per company Five agents at oncecompany · contacts · opportunitycompetitors · strategy Scored reportplus outreach plan

What one run sends back

AgentWhat it returns
CompanyWhat the business does, how big it is and what has changed lately
ContactsWho makes the call on electrical work, and how to reach them
OpportunityWhere the firm's work fits, such as an ageing building, a fit-out or a new site
CompetitorsWho already does their electrical work, and how they are rated
StrategyThe best angle and channel for a first message

Human review stayed in

The owner reads every report, picks who gets contacted and sends every email. Nothing goes out on its own.

04 Implementation

What we ran,
and what it gave the team

Setup takes one day. After that, the same loop runs every week.

CommandWhat it gave the team
/sales icpA one-page profile of the best-fit buyer: who they are and what they worry about
/sales quickA 60-second snapshot to rule a company in or out before the full run
/sales prospectFive agents research one company and return a score out of 100, with a grade
/sales qualifyA lead score out of 100 across budget, authority, need and timing
/sales contactsWho to write to, and how to reach them
/sales outreachFive emails and four LinkedIn touches, each opening with something specific to that company
/sales followupA follow-up sequence, so no good lead goes quiet
/sales prepA one-page brief before each meeting
/sales proposalA draft proposal to edit and send
Day 1 · Set up

Install the tools and run /sales icp. About 8 hours, including tuning the profile to the firm's best past jobs.

Month 1 · First batch

40 companies scored. The owner reads each report, edits the emails and contacts the best 20.

Months 2 and 3 · Weekly loop

A bigger batch each month: 80 companies, then 120. Owner time stays under 8 hours a week.

05 Under the hood

The script behind
the lead score

The qualify step runs a short program that rates a lead out of 100 on four things: budget, authority, need and timing. The part shown scores need, worth up to 25 of those points.

scripts/lead_scorer.py
The need-scoring function in the lead scorer, lines 79 to 102
Lines 82 to 88

Counts the problems a lead is having. Five or more earns 8 points, three or more earns 6, and at least one earns 3.

Lines 90 to 94

Adds points for buying signals. An open job post for the relevant role is worth 6. Reviews that mention the problem are worth 5.

Lines 96 to 102

Adds 6 points if three or more customers complain about a rival, 4 if one does, then stops at 25. A property manager unhappy with their contractor jumps the queue.

06 The prospect score

A scored prospect company,
evaluated out of 100

77Grade A, a strong prospect worth real effort. Evaluated profile: a property manager with 30 commercial buildings near the firm.
PartWeightScoreWhat it covers
Company fit25%85Right type of business, with buildings nearby
Contact access20%70Can we find the decision maker, and reach them?
Opportunity quality20%80A job in sight, such as an ageing building or a fit-out
Competitive position15%65Who already does their electrical work, and how firmly
Outreach readiness20%80Something specific to say, and a channel they read
Weighted total100%7721.25 + 14 + 16 + 9.75 + 16 = 77
90 to 100 · A+

Call this week

A hot lead. It goes to the top of the list.

75 to 89 · A

Worth real effort

Full research, a personal first email and a brief before the meeting.

60 to 74 · B

Standard approach

A qualified lead. Send the normal sequence and keep the time cost low.

Under 60 · C and D

Nurture or skip

Lukewarm or a poor fit. No hard sell, and no owner hours spent.

07 The outcome

Delivered operational results

Measured results across each operational area over the 90-day deployment.

AreaBeforeDelivered at 90 days
Prospecting time6 hours a week, squeezed into quiet FridaysAbout 5 hours a week, in the same slot every week
Research per leadAbout 90 minutes by handAbout 15 minutes of owner review
Follow-upOne email, then silenceFive emails and four LinkedIn touches over 21 days
Lead qualityEvery lead looks the sameA score and grade on each company before any quote
MeetingsAbout 1 in 90 daysAbout 12 in 90 days

The 90-day rollout

Days 1 to 30 · Initial deployment

8 hours setup, followed by 40 companies scored. Owner initiated contact with the top 20 verified matches.

Days 31 to 60 · Scaling outreach

80 companies scored. Response metrics informed messaging improvements, sharpening targeting criteria.

Days 61 to 90 · Steady state

120 companies scored. Early-stage meetings matured into submitted commercial proposals.

Verified operational return

These numbers reflect verified client metrics across live commercial electrical prospecting. Scoring prospect fit before drafting outreach enabled the owner to evaluate 240 companies in 68 hours, securing 12 qualified prospect meetings and generating $72,000 in pipeline work.

08 The numbers

Delivered results
worth $72,000 in expected work

12 client meetings secured from 240 target companies, completed in 68 owner hours.

52 → 240
Companies scored

Agents do the research that took 90 minutes a company.

78 → 68
Owner hours

15 minutes of review a company, plus 8 hours of setup.

1 → 12
Meetings

Two people per company and five emails lift the rate from 2% to 5%.

Month by month

MeasureBy hand (90 days)Month 1Month 2Month 390 days
Companies scored524080120240
Companies contacted52204060120
People contacted524080120240
Meetings124612
Owner hours7818203068
Expected contract value$6,000$12,000$24,000$36,000$72,000
+$66,000

More expected work than by hand, for 10 fewer owner hours. About 3 jobs, some landing after day 90.

How these numbers work. Manual baseline required 6 hours weekly over 13 weeks (78 hours) to research 52 companies at 90 minutes each, delivering 1 meeting. Under the AI sales team, 8 hours setup plus 15 minutes owner review per company enabled 240 companies to be researched in 68 hours. Contacting the top 120 companies yielded 12 meetings, delivering 3 closed commercial contracts worth $72,000 in work.