Rickey SinghAn entrepreneur & AI transformation strategist
Apply

Use case · Home services · 7-day AI sprint readout

Where AI will make money at Northfield in the next 90 days.

Seven days of interviews, call data, quote history and job records, reduced to one decision: three AI projects worth about $1.04M a year in gross profit, the order to build them in, and the rules that keep them safe.

Client
Northfield Home Services
Business
HVAC, plumbing and electrical, residential
Size
140 people · $31.4M revenue
Prepared by
Rickey Singh
Sprint
7 days · 2 leadership sessions · 9 interviews

Use case scenario: the company name and figures are representative of a typical engagement, so no client data is shared. Client reports stay confidential.

01 · Executive summary

Northfield is losing money at the phone and after the quote, not in the field.

The field team is strong, and close to capacity. The leak is upstream: one call in seven goes unanswered, and most replacement quotes get no follow-up after day three. Both are problems AI handles well today, with the tools Northfield already pays for.

$1.6MAnnual gross profit across all 14 opportunities found
$1.04MFrom the top three alone, 65% of the total
90 daysTo have the top three live and measured
~5 wksExpected payback once all three are live

Decision agreed at the readout

Start quote follow-up on Monday. Pilot the AI receptionist after hours from week 3. Job notes follow in week 7.

Owners are named, measures are agreed, and the leadership team reviews results at day 90 to keep, change or stop each project.

What we found

  • 13,440 calls a year go unanswered, 14% of inbound volume, mostly evenings, weekends and Monday mornings.
  • 58% of open replacement quotes receive no contact after day three. The close rate is 34%.
  • Technicians spend 41 minutes a day on job notes, photos and invoice detail, mostly after the last job.
  • Staff use personal AI accounts with customer names in them. No policy exists yet.

What we recommend

  • AI quote follow-up, drafted by AI and approved by advisors, inside the CRM.
  • An AI receptionist for after-hours and overflow calls that books jobs into the dispatch board.
  • Voice-to-job-notes for technicians, producing the invoice detail and the customer summary.
  • A one-page AI policy and company AI accounts before anything goes live.

02 · AI readiness assessment

Ready for the top three, with two fixes first.

Readiness was assessed only where it affects the recommended projects. Click a row to see the evidence and what has to change.

Verdict: the data and tools are good enough to start now. Governance and ownership are the gaps, and both are fixed in the first two weeks of the plan.

03 · Opportunity map

14 ideas, scored on return, effort and risk.

Every candidate came from the interviews and the data. Each one is scored on annual gross profit, the effort to launch and run it, and the main risk. Hover a point for detail; filter the table to see each group.

Annual gross profit vs. effort to launch

#Opportunity Verdict

Gross profit uses Northfield's own margins: 48% on service work, 42% on installs. Time savings are valued at a $48 loaded hourly cost. Quick wins are left to Northfield's own team.

04 · The top three

What to build, why, and how we will know it worked.

Each project runs inside tools Northfield already uses. Open "Show the math" to see how each value was calculated.

1

AI follow-up on replacement quotes

Owner: Sales Manager · Live in week 2 · Effort: low · Risk: low

$426K
Gross profit / yr

The problem

Northfield wrote 2,150 replacement quotes last year and closed 34%. Quotes that got a call or message in the first week closed at 46%; the rest closed at 27%. Advisors are in homes all day, so follow-up slips.

What we build

  • AI drafts a personal follow-up from the advisor's visit notes, photos and the quote: day 1, 3, 7 and 14.
  • The advisor approves each message from their phone in one tap.
  • Replies that mention price or timing go straight to the advisor.

Measure and guardrail

  • Measure: close rate from 34% to 38% within 90 days.
  • Guardrail: AI never changes a price or offers a discount; the first 200 messages are checked by the Sales Manager.
Show the math
Replacement quotes a year2,150
Close-rate lift (34% → 38%)+4 pts
Extra installs a year86
Average install value$11,800
Extra revenue$1,014,800
Gross profit at 42% margin$426,000
2

AI receptionist for after-hours and overflow calls

Owner: Customer Service Manager · Pilot in week 3 · Effort: low · Risk: medium

$325K
Gross profit / yr

The problem

Of 96,000 inbound calls, 13,440 went unanswered. Two in five were from people trying to book work. After hours, callers reach an answering service that takes a message; a third of those callers have booked elsewhere by morning.

What we build

  • An AI voice agent answers when no one is free, says it is an AI, and books the job into the dispatch board.
  • It knows the service area, the published call-out fee and the membership terms.
  • Emergencies (gas smell, flooding, no heat below freezing) transfer to the on-call technician at once.

Measure and guardrail

  • Measure: at least 30% of would-be-missed booking calls become booked jobs.
  • Guardrail: no price beyond the published fee; any caller can reach a person; 20 calls a week are reviewed for accuracy.
Show the math
Unanswered calls a year13,440
Share that were booking calls40% → 5,376
Recovered by the AI receptionist (30%)1,613 jobs
Average service ticket$420
Extra revenue$677,500
Gross profit at 48% margin$325,000
3

Voice-to-job-notes for technicians

Owner: Field Operations Manager · Pilot in week 7 · Effort: moderate · Risk: medium

$285K
Value of time / yr

The problem

Ride-alongs and system data show technicians spend 41 minutes a day typing notes, attaching photos and itemising invoices. Notes are thin, so the office calls technicians back with questions, and warranty claims stall.

What we build

  • The technician talks for 60 seconds at the end of a job; AI writes the job notes, the itemised invoice lines and a plain-English summary for the customer.
  • Photos are labelled and attached automatically.
  • The technician checks and submits.

Measure and guardrail

  • Measure: admin time down by 25 minutes a day per technician, and office call-backs down by half.
  • Guardrail: the technician approves every invoice; parts and prices come from the price book, never from the AI.
Show the math
Technicians62
Time saved per technician per day25 min
Working days a year230
Hours returned a year5,942
Value at $48 loaded hourly cost$285,000

05 · 90-day plan

One project at a time, each one proven before the next.

Projects are staggered so the team learns on the lowest-risk one first. Each has a single owner and a measure checked every week.

Quote close rate
34%→38%

Checked weekly from the CRM. Owner: Sales Manager.

Missed booking calls recovered
0%→30%+

Checked weekly from call logs. Owner: Customer Service Manager.

Technician admin time / day
41m→16m

Checked in the pilot group first. Owner: Field Operations Manager.

06 · The AI operating system at day 90

One place where the leadership team sees every AI worker, and every result.

The three projects run inside Northfield's existing tools, and they report to one place: an AI operating system the owners open each morning. It shows what each AI worker did overnight, what is waiting for a person to approve, and whether the numbers are moving.

Northfield · AI operating system

Good morning, Dana

Week 13 of the 90-day plan · all guardrail checks passing

TodayThis week90 days
Calls answered by AI
248
was 0 · missed before
Jobs booked by AI
33
+$13.9K revenue
Quote close rate
37.6%
▲ from 34.0%
Tech admin / day
18m
▼ from 41m

Quote close rate, weekly target 38%

Receptionist248 calls · 96% handled · 9 handed to a person
Quote follow-up61 messages · 100% approved before sending
Job notes41 jobs · 38 approved, 3 waiting

Waiting for you 3 to approve

Quote #4821 · Day 3 follow-upMike R.

"Hi Laura, following up on the heat pump quote from Tuesday. Happy to walk through the rebate…"

Invoice · Job 18-2207Carlos T.

Water heater flush, anode rod (price book), 1.5 h labour. Summary for customer ready.

Quote #4834 · Day 1 thank-youPriya S.

"Thanks for having us out today, Mr. Okafor. Your quote for the 3-ton system is attached…"

Overnight activity live

☎
Receptionist booked a no-cooling call · Westbrook · tomorrow 8–10am · told the caller it is an AI
!
Gas smell reported, transferred to on-call tech · 14 seconds to a person · emergency rule
✉
Quote follow-up drafted for 6 open quotes · waiting for advisor approval
✎
Job notes written for 41 jobs · 38 invoices approved by technicians, 3 waiting

Four screens from the system the owners open each morning: the overview, the AI receptionist's calls, the quote follow-up pipeline, and technician job notes turned into invoices. Swipe or use the arrows. Every figure matches the rest of this use case.

07 · Guardrails

Rules that protect customers without slowing the team down.

These go into a one-page AI policy signed off by the leadership team in week 1, before any project goes live.

Data that AI may use

Job history, quotes, call recordings and the price book, through company accounts only. Never card numbers, and no personal AI accounts with customer data.

Where a person approves

Every quote message until the error rate stays under 1% for four weeks, every invoice, and any change to price, discount or warranty.

How accuracy is checked

A weekly sample of 20 AI calls and 20 AI messages, scored against a checklist. Two misses in a week pauses the project until fixed.

Safety first on calls

The AI says it is an AI. Emergencies and upset callers go to a person at once. Any caller can ask for a person and get one.

08 · Costs and payback

$1.04M of gross profit, run on software Northfield mostly already licenses.

Build costs are one-off and fixed, scoped for each project after the consultation. Running costs are software and usage, billed by the vendors at cost.

What it costs to run

Software and usage, top three$3,600 a month
Running costs, year one$43,200
Build costsFixed price, after consultation

Return vs. cost, year one

Profit · $1.04M
Running cost · $43K

Running costs are about 4% of the expected profit in year one. Even at half the expected result, the top three pay for themselves within months, which is why the plan is safe to start.

09 · Next step

Three ways to run the plan.

Option 1

Your team runs it

This report says what to do. The Implementation Playbook says how.

Everything above is written for Northfield's managers to deliver themselves, with a check-in at day 30. Teams that run it in-house can add an Implementation Playbook with step-by-step build guides, prompts, vendor setup and test checklists, scoped to their projects after the consultation.

Option 2

Rickey builds the first win

Fixed scope, fixed timeline, handed over ready to run.

Rickey builds the opportunities you choose from your report, inside the tools your team already uses. Each one is tested against its measure and handed over with the team trained to run it.

Option 3

Fractional AI leadership

A monthly retainer, 6-month minimum.

Rickey owns the roadmap as fractional Chief AI Officer, runs the builds and the AI workers, and reports results to the leadership team monthly.

Option 1 · how it works

Your team runs it

Best for a business with an operations or IT lead who has time to own delivery.

  1. You receive the reportThe ranked opportunities, the 90-day plan with an owner for every workstream, the measures and the guardrails, as in this use case.
  2. Your managers own the workEach owner follows the week-by-week plan in section 05 and tracks the measures their project commits to.
  3. Add the Implementation Playbook if you want the howStep-by-step build guides, prompts, vendor setup and test checklists for each project. It is scoped to your projects after the consultation.
  4. Day-30 check-in with RickeyA review of the measures against the plan: what is working, what is stuck, and what to change for days 31 to 90.

You can move to option 2 or 3 at any point.

Apply for a private consultation

Option 2 · how it works

Rickey builds the first win

Best for a business that wants a first result in weeks, without hiring.

  1. Choose your buildsTick the builds below that interest you. After the consultation, your report gives every recommended build a fixed scope, timeframe and price, and you confirm the ones you want.
  2. Week 1: kick-offAccess to the tools you already use, the measure each build must move, and the guardrails it must follow, agreed in writing.
  3. Build and testEach build runs inside your existing systems, with a person approving its output until it meets the quality bar in your guardrails.
  4. Go live and hand overLaunch, team training and a written runbook, so your people run it without depending on Rickey.

Build list

Pick your builds

Every opportunity from this report. The top 3 are ticked to start; add or remove any.

Selected0 builds
Delivery–
Apply for a private consultation

Delivery time is an estimate. Scope and price for each build come in your report after the consultation. You can start with the top-ranked build and add more later.

Option 3 · how it works

Fractional AI leadership

Best for a business that wants one senior owner for AI, without a full-time executive salary.

Monthly retainerFrom $6,000Set by days a month and how much delivery is included.
Commitment6 monthsThe shortest period in which the roadmap and first builds show results.
Client load2 to 3Rickey works with two or three businesses at a time, by design.
  1. Consultation and scopeIt starts with the one-hour Private Executive Consultation. If fractional leadership fits, the days a month, the builds included and the fixed monthly fee are agreed in writing.
  2. Days 1 to 30: the roadmap and the first winThe 7-day sprint report becomes a 12-month roadmap, the guardrails go in, and the highest-return build goes live inside your existing tools.
  3. Every month: run, build and reportRickey owns the roadmap, runs the builds and the AI workers, chooses vendors on merit, and trains your team. The leadership team gets a monthly report on spend, what shipped and what it returned.
  4. Month 6: review and decideMeasures against the plan. You continue, scale up, or hand over to your team with runbooks for every AI workflow.

Builds beyond the agreed scope are quoted at fixed prices. Software is billed by the vendors at cost.

Apply for a private consultation

Your numbers, your plan

Want this report for your business in 7 days?

Every sprint starts with a one-hour Private Executive Consultation. You leave it knowing where AI could pay back in your business and whether a sprint is the right next step.

Apply for a private consultation How the 7-day AI sprint works

Appendix · Method and assumptions

Sources in a real sprint: 12 months of call logs, CRM quote history, field-service job records, the price book, two leadership sessions and nine short interviews with customer service, sales, dispatch and technicians.

Scoring: return is annual gross profit at the business's own margins, or time valued at loaded cost. Effort runs from 1 (configure an existing tool) to 5 (a new system with integrations). Risk covers customer impact, data sensitivity and accuracy.

About this use case: Northfield Home Services is a representative scenario, not a named client. The numbers are typical for a business of this size and sector. Real client reports are confidential.