The Job Loop: Catch, Book, Collect, Keep, plus Run and Prove
Learn the framework every lesson uses: Catch, Book, Collect and Keep, with Run underneath and Prove around it.
Read the lesson
Goal: Place any problem an owner describes into one Job Loop stage. Name the metric and the ServiceAgent tool that fix it.
You need: Lessons 1 and 2. Your practice account from lesson 1.
Why this matters
Owners rarely name their leaks clearly. They say "the phones are crazy" or "we are too busy to follow up". These are feelings, not problems you can fix. If you jump straight to features, you build the wrong thing first.
The Job Loop turns a vague complaint into four facts: the stage, the metric, the fix and the package. Every course in SAU uses it. Every lesson you build later belongs to one stage. When you know the map, you can audit, sell and report with the same words.
The map also keeps you honest. You fix the stage where the job stops moving first. A business that misses 20% of its calls gains little from better review requests.
The four stages
Every service job moves through four stages, in order:
- Catch. Answer every call, text, chat and form. The job exists only if someone catches it.
- Book. Put the job on the calendar and get the customer there.
- Collect. Close the quote and get paid.
- Keep. Get the review, sell the plan and bring the customer back.
The Job Loop
- Catch
- Book
- Collect
- Keep
Each stage has its own typical leak and its own measure.
| Stage | Typical leak | Measure to watch | Main ServiceAgent tools |
|---|---|---|---|
| Catch | Missed calls, slow replies to web leads | Answer rate | Voice Agents, Conversations, Widgets |
| Book | No-shows, double-booking, empty slots | No-show rate | Jobs, booking page, Workflows reminders |
| Collect | Estimates that go cold, slow payment | Estimate close rate, days to paid | Payments, Contacts → Deals |
| Keep | Few reviews, no plans, no repeat work | Reviews per month, renewal rate | Review requests, Surveys, Referrals & loyalty |
The numbers behind each leak are real. Call-tracking data puts missed calls at 14% to 27%. Only 37% of companies answer a web lead within an hour. Only 37% of in-home sales close on the first visit. And 97% of consumers read reviews before they pick a local business.
The leaks behind each stage
What good looks like in each stage
Catch. Every call gets answered, day and night, by a person or the AI voice agent. Every missed call gets a text back within a minute. Every web form and chat gets a reply in under 60 seconds. For an owner-operator, this means the phone stops pulling them off jobs. For a fleet, it means peak days do not overwhelm the office. Keep the 10-call limit in mind: the AI takes up to 10 calls at the same time.
Book. The customer can book by phone, text or the booking page. The calendar knows how long each job really takes. Buffers leave time to drive. Reminders go out 24 hours and 1 hour before the visit. For a fleet, each job has one person in Assigned To, and vehicles are tracked as resources.
Collect. Every estimate gets follow-up on a schedule, not when someone remembers. The customer can pay by card from a text link. Overdue invoices get a polite nudge from a workflow. The owner sets every price. The Operator builds the structure around it.
Keep. Every customer gets a review request after the job. Every review gets a reply. Plan members get their tune-ups on time. Customers who have not booked in a year hear from the business again, if they gave consent.
A business rarely needs all of this on day one. Use the list to see the gap between today and good.
Run and Prove: the two layers
Two layers sit across all four stages.
Run is the back office. It is the CRM, the tasks, the workflows and the AI agents. Run keeps the business moving without the owner holding it in their head. If callbacks and follow-ups live only in the owner's memory, that is a Run leak. No system tracks them, so they slip.
Prove is measurement. It shows each stage's number every month. Without Prove, the owner cannot see what changed, and you cannot show your value. Owners often cannot say which channel produces booked jobs. That is a Prove leak.
Run is also where the business memory lives. Before ServiceAgent, prices, service areas and policies sit in the owner's head and phone. After setup, they live in the AI knowledge base, the CRM and the call transcripts. When a tech leaves or the owner takes a holiday, the business keeps working.
Where the business memory lives
- Prices in the owner's head
- Service areas on their phone
- Callbacks remembered, or not
- Prices and policies in the AI knowledge base
- Customers and deals in the CRM
- Every call in a transcript
- Follow-ups as tasks and workflows
Prove uses the same six numbers every month for every client. You learn them in Prove: Reporting and the Monthly ROI Review. For a multi-location brand, Prove also compares locations. The spread between the best and worst location often shows the next fix.
Run and Prove are not extra stages a job passes through. They support every stage. A missed-call text-back is a Catch fix built in Run. A monthly answer-rate report is Prove for Catch.
How to place a leak
Use one rule: a leak sits in the first stage where the job stops moving.
Ask the owner to walk you through a lost job. Then ask one question per stage:
- Catch: Did we hear from the customer, and did we reply fast?
- Book: Did the job reach the calendar, and did the customer show up?
- Collect: Did the customer say yes to the quote, and did we get paid on time?
- Keep: Did we ask for a review, and did we contact the customer again?
The first "no" marks the stage. Then ask a Run question: is this tracked anywhere, or only in someone's head? Then ask a Prove question: would we have known about it without this conversation?
Place a leak with the first-stop rule
- Walk through one lost job with the owner
- Catch: did we hear from the customer and reply fast?YesGo on to BookNoCatch leakEnds here
- Book: did the job reach the calendar, and did the customer show?YesGo on to CollectNoBook leakEnds here
- Collect: did the customer say yes, and did we get paid on time?YesGo on to KeepNoCollect leakEnds here
- Keep: did we ask for a review and contact the customer again?YesNo stage leak foundNoKeep leak
- Run: is it tracked anywhere? Prove: would we have known?
Size changes where leaks sit
The stages are the same at every size. The weak stage often differs.
| Size | Common weak stage | Why |
|---|---|---|
| Owner-operator | Catch | The owner is on the job when the phone rings |
| Fleet, about 20 techs | Book and Collect | Gaps idle a crew; sales reps follow up unevenly |
| Multi-location | Run and Prove | Each location works differently; head office cannot compare them |
Treat this as a starting guess. The audit tells you the truth for each client. A multi-location brand can still miss calls at one site. A solo owner can still lose big quotes.
The packages follow the loop
SAU's three packages add stages in order:
- Never Miss a Job covers Catch.
- Business on Autopilot covers Catch, Book and Collect.
- Growth Engine covers all stages, including review requests and replies in Keep.
The packages follow the loop
- 1Never Miss a JobCatch
- 2Business on AutopilotCatch, Book and Collect
- 3Growth EngineAll stages, including review requests and replies in Keep
You usually start with Catch, prove it, then add one stage at a time. You learn this in Package, Price and Deliver.
Build it in ServiceAgent
Map the app to the loop in your practice account. This takes about 10 minutes.
- Open
Voice Agents. Write "Catch" next to it. - Open
Conversations. This is the inbox for texts, email and chat. Write "Catch". - Open
Widgets. FindChat & phone(Catch) andBooking calendar(Book). - Open
Jobs. Check the viewsDay,Week,MonthandList. Write "Book". - Open
Payments. Check the tabsInvoicesandPayment links. Write "Collect". - Open
Contacts, then theDealstab. Estimates move through the pipeline here. Write "Collect". - Type "Review requests" into
Search or run a command.... Open the page. Write "Keep". - Open
Surveys. Write "Keep". - Open
WorkflowsandAsk Emma. Write "Run". - Open
Dashboard. Then type "Analytics" in the search box and open it. Write "Prove". - Check it works: you have a one-page map with every main screen tagged to a stage.
Worked example: Polar Air Services
Polar Air Services is an HVAC fleet of about 20 techs on Growth. In a first meeting, the owner lists five complaints. You place each one:
| Owner says | First stage where the job stops | Why |
|---|---|---|
| "On hot days, the phones ring off the hook." | Catch | Calls go unanswered |
| "Techs show up and nobody is home." | Book | The customer did not show |
| "We quote lots of new systems. Few close." | Collect | Estimates go cold |
| "Our reviews are from three years ago." | Keep | Nobody asks |
| "I don't know which ads work." | Prove | No measure links ads to jobs |
Now decide what to fix first. Say Polar gets about 50 calls a business day and misses 20% on peak days. In a month that is 50 × 22 = 1,100 calls and 220 missed calls.
Say the AI agent lifts the answer rate so only 5% are missed. That is 55 missed calls, so 165 more calls get answered. At the 42% booking rate, about 69 more jobs book. At an example $400 ticket, that is about $27,700 a month.
No Book, Collect or Keep fix moves that many jobs this fast. So you start with Catch. Then you add Book, because more booked jobs mean more chances for no-shows.
Polar Air Services: fixing Catch first
Common mistakes
- Fixing the loudest complaint first. Fix the first stage where jobs stop. Upstream fixes feed every later stage.
- Calling Run a stage. Run is the back office under every stage. A follow-up that lives in the owner's head is a Run leak.
- Building all stages at once. You cannot tell what worked. Add one stage, measure it, then add the next.
- Skipping Prove. If you do not measure the stage before you change it, you cannot show the owner the result.
- Treating size as a rule. A small shop can have a Collect leak, and a fleet can miss calls. Let the audit decide, not the size table.
Do this now
Interview a business owner you know for 15 minutes. If you cannot, use a review site. Read 10 of a local company's 1-star and 2-star reviews. Place each complaint in a Job Loop stage, using the first-stop rule. Count the complaints per stage. Write the stage you would fix first and one sentence on why. Then name the package that covers it. Keep the page; you use it in The Operations Audit.
Practice the conversation
- Owner Honestly, everything is a mess. I don't know where to start.
- Operator That's normal. Let's follow one lost job from the first call to the end.
- Owner OK. Last week a lady called about their AC at 6 pm. Nobody picked up.
- Operator Then the job stopped at the first step. We call that Catch.
- Owner And what about the quotes that never close?
- Operator That is Collect. It matters too, but a quote needs a caught call first.
- Owner So you fix the phones first?
- Operator Yes. We prove it works in 30 days. Then we fix the next stage.
- Owner How will I know it worked?
- Operator You'll see your answer rate and booked jobs each week, in your own numbers.
Quick check
- At an HVAC shop, callbacks and follow-ups live only in the owner's head. No system tracks them. Which part of the Job Loop holds this leak?
- Catch
- Run
- Keep
- Prove
- A customer called, booked and had the work done. The invoice is now 45 days unpaid. Where does this job stop?
- Catch
- Book
- Collect
- Keep
- A plumber gets 30 calls a business day and misses 20%. You cut misses to 10%. Use 22 business days and a 42% booking rate. About how many more jobs book each month?
- 28
- 66
- 132
- 277
ServiceAgent UniversityLearn to run a service business on autopilot. Free courses, Blueprints and certification for Service Operators.