HVAC Business Starter Kit Review: I Re-Priced All Five Jobs. The Math Held.
The short version: the HVAC Business Starter Kit claims a labor-rate build-up, a ten-line job formula, three estimate templates and a startup cost model with a self-check. The arithmetic held — I re-implemented every published figure from scratch and all five worked jobs, the seven-line rate chain, the five-rung truck ladder, the tracker totals and the agreement math came back identical to the cent. The self-check did not hold: self_qos.py is a shipped-sample integrity test, not a validator for your own numbers, and it refuses your edits with a message that reads like corruption.
The arithmetic held. The self-check did not.
The problem this thing is trying to solve
Ask three HVAC contractors what they charge for a service call and you get three numbers and no reasoning. Ask them how they got that number and you get the honest answer: "that's about what the shop down the road charges," or "forty years of doing it." Neither is a method. It is a number copied from a business with different insurance, different fuel costs, different billable hours and a different appetite for risk.
The failure is not that the number is too low. It is that nobody can tell which it is. A rate picked because it sounds fair means a week booked solid can still lose money, and you find out at month end when the account is empty and the calendar was full. The same disease shows up in the parts line — a capacitor marked up "a bit," an after-hours call priced from memory, a first-time customer discount invented on the driveway — and none of those decisions leave a trace you can compare against next quarter.
The printable version of the fix is unglamorous: one rate built from your own cost base, and one fixed order of operations applied to every job so that two estimates for the same work cannot disagree.
A note on scope, because this kit has been written about on this blog before. The earlier piece walks the method the way the kit intends it to be walked: the rate build-up explained, one job stepped through line by line, the break-even framed. That is the right way to learn the method. This piece is the adversarial follow-up — I re-implemented every formula independently, re-priced all five jobs from my own code, and pushed the bundled checker until it broke. If you want the teaching walk-through, read the earlier article. If you want to know whether the arithmetic is real and where the guard rails stop, this is that part.
What is actually in the box — counted, not read off the sales page
I unzipped the download into an empty directory and counted the files rather than reading the listing:
- 17 files, 85,815 bytes unpacked, 39,265 bytes as the ZIP you download.
- 13 Markdown documents, 2 CSVs, 1 Python script, 1 licence file.
self_qos.py— 196 lines, standard library only. Imports:csv,os,re,sys,decimal. No network, no dependencies.- The two CSVs are the working models: an 18-row startup cost ladder (base costs plus per-truck costs, low and high band each) and a 6-row job cost tracker.
No accounts, no software, no plugins, no monthly fee. A text editor and Python 3. That is the whole delivery.
What held: the arithmetic, recomputed from scratch
I did not trust the documentation. I wrote my own implementation of the published formulas using Python's decimal module with explicit ROUND_HALF_UP, then fed it the same inputs the kit ships and compared against the frozen figures in the files.
1. The seven-line labor-rate chain reproduces exactly
The example starts from a $38.00 wage and lands on an hourly rate. Every intermediate froze at the documented value:
| Line | Kit claims | What I computed |
|---|---|---|
| payroll_load | 8.36 | 8.36 |
| loaded_labor | 46.36 | 46.36 |
| labor_cost_month | 7,417.60 | 7,417.60 |
| labor_cost_hour | 57.95 | 57.95 |
| overhead_hour | 25.35 | 25.35 |
| cost_hour | 83.30 | 83.30 |
| labor_rate | 104.13 | 104.13 |
That last line is the interesting one. 83.30 ÷ 0.80 = 104.125, which is an exact tie on the half cent. The kit documents that it rounds half-up to 104.13 and tells you to copy the figure rather than recompute it with a different rule. That instruction is correct and it matters — see the first leak below.
2. All five worked jobs reproduce, line by line
Five jobs are priced end to end from the same ten-line formula. I re-implemented the formula and re-priced all five. Total, job cost and margin all matched, on all five rows:
| Job | Type | Kit total | My total | Kit cost | My cost |
|---|---|---|---|---|---|
| J-001 | Furnace tune, routine call | 281.95 | 281.95 | 121.98 | 121.98 |
| J-002 | Capacitor, first-time customer | 409.65 | 409.65 | 245.00 | 245.00 |
| J-003 | After-hours emergency | 1,180.49 | 1,180.49 | 564.66 | 564.66 |
| J-004 | Three-ton changeout with agreement | 8,850.90 | 8,850.90 | 6,460.20 | 6,460.20 |
| J-005 | Holiday emergency, both premiums | 962.61 | 962.61 | 388.30 | 388.30 |
That is not a small thing. The usual failure in a template bundle is that the worked example does not survive being re-priced — the discount is applied to the wrong base, the premiums multiply instead of compose, the trip charge gets a percentage on top. Here the four ordering rules the kit states are the ones its own numbers obey: premiums applied once on the single adjusted line, trip charge never marked up, agreement credit on labor only, first-time discount on materials only.
3. The truck ladder and the tracker totals add up
- The 18 cost rows split into 5,290.00–16,610.00 of base cost and 7,100.00–24,400.00 per truck. One truck: 12,390.00–41,010.00. Five trucks: 40,790.00–138,610.00. Both match the files.
- The 6-row tracker sums to 11,780.60 revenue / 7,851.50 cost / 3,929.10 margin. Every row's revenue-minus-cost equals its stated margin.
- The agreement model: 190.00 revenue per system per year, 142.72 cost, 47.28 gross. At 50 customers that is 9,500.00 / 7,136.00 / 2,364.00, which matches the table.
What leaked — two, both real
Leak one: the self-check validates the sample, not your workbook
The README tells you to edit 01_Startup_Cost_Model.csv to match your own truck count and local costs, and it is honest that the ladder line will then report SELF-QOS FAIL "because your totals now differ from the example." That is a fair warning and I credit it. What the README does not say is how far the failure reaches.
self_qos.py opens the CSV and asserts that it contains exactly 18 data rows, that each low figure is above zero and each high figure at or above its low, and that the two column sums produce one specific five-number ladder. Those are integrity checks on the shipped sample. They are not validators for a business model you filled in. I ran four edits a real purchaser would make:
| Edit you would make | self_qos.py says |
|---|---|
| Lower one line's high figure inside its valid band | FAIL - ladder for 1 truck(s) is 12390.00/40010.00, expected 12390.00/41010.00 |
| Add one honest cost the kit omits (shop rent) | FAIL - startup model has 19 data rows, expected 18 |
| Remove a line that does not apply to you | FAIL - startup model has 17 data rows, expected 18 |
| Add a per-truck line for a second van | FAIL - startup model has 19 data rows, expected 18 |
Every one of those is a legitimate edit, and every one is reported with the word FAIL and a message that reads like corruption. The row-count message is especially misleading: it calls a 19-row file, with no invalid values in it, a failure. The script's own source confirms the cause — the rate chain check in check_rate_chain() has the wage, payroll load, paid hours, billable hours, overhead and margin hardcoded as literals (38.00, 0.22, 160, 128, 3245.00, 0.20). It never reads your inputs. Drop your own wage into the pricing guide, exactly as the guide instructs, and self_qos.py still prints PASS: it is comparing the constants against themselves.
So the tool does exactly one useful thing: it proves your download is intact and its frozen example is internally consistent. Run it once on the untouched kit, keep the PASS line as your receipt, then stop running it. Treating it as a green light on your own cost model is the mistake it invites.
Leak two: the rate is a tie-breaking rule, and Python's default gets it wrong
The kit is unusually careful here — it names the four values in its example that sit exactly on a half cent and says to copy them rather than recompute with a different rounding rule. It is right to be careful, because the default rounding a Python user would reach for disagrees:
round(83.30 / 0.80, 2) -> 104.12 # banker's rounding ROUND_HALF_UP to 104.13 -> 104.13 # the kit's rule
That one-cent difference is not cosmetic. Re-price all five jobs at the banker's-rounded rate and every one of them moves: J-001 by −0.02, J-002 by −0.01, J-003 by −0.04, J-004 by −0.12, J-005 by −0.04. Individually trivial. Collectively they are the reason two people using the same kit can hand the same customer different prices and both be "following the method." If you re-implement the formula in a spreadsheet or your own script — which the plain-file format invites — you must specify half-up explicitly. Excel's ROUND is half-up and matches; Python's built-in round(), JavaScript's toFixed in some paths, and any IEEE-754 default do not.
Neither leak is a defect that makes the kit wrong. Both are places where the documentation promises more protection than the code delivers, and both are cheap to work around once you know.
Who it is NOT for
- Not for you if you have not finished your certification or your hours. The kit's first cost line is the EPA 608 exam; the pricing model assumes you already have the credential and are pricing work, not studying for it. It will not help you get certified and it does not teach service technique.
- Not for anyone who wants a software suite. This ships no dispatch board, no scheduling app, no invoicing, no CRM, no maps, no API and no integration with any field-service platform. It is Markdown, two CSVs and a 196-line checker. If you expect a dashboard, you will be disappointed in the first five minutes.
- Not for anyone reading it as a price list. Every number in it is an example input — a $38.00 sample wage, a $3,245.00 sample overhead. None of them is a market rate, a regional average or a benchmark. Copying the 281.95 service call because it is printed in the file is exactly the guess-a-number behaviour the kit exists to replace.
- Not for anyone looking for licensing, permitting, insurance or code guidance. The compliance checklist tells you which questions to ask and where the risk sits. It is not advice and it does not know your state's rules. Verify every compliance line against your own authority — the kit says so twice and means it.
- Not for a team that wants the rules enforced automatically. There is no linter and no CI gate. The arithmetic check is a sample integrity test, as measured above. Enforcement is you, running the numbers.
- Not for a business already running priced work with a system that reconciles. If your estimates and your month-end already agree, this is a rewrite, not a fix.
Honesty block
Example figures are illustrative planning inputs — not market rates and not earnings. Nothing here is legal, tax, insurance, licensing, or financial advice; HVAC work carries refrigerant-handling, electrical and gas rules that vary by jurisdiction and that this kit does not interpret for you. No claims are made about CRM, map, API, invoicing, dispatch or scheduling integrations, because the bundle ships none — it ships 13 Markdown files, 2 CSVs and one 196-line Python checker, as counted above. No ratings, reviews, testimonials or sales figures are quoted in this article, because I have none that would be honest to quote: the live listing shows a rating count of zero and no sales.
Price and the link
Disclosure: this post promotes a product we made. The sale is ours directly — this is not an affiliate link and there is no third-party commission on it. I am writing it because the arithmetic above is worth having in public before anyone spends money, not because a commission depends on it.
The kit is on Gumroad at https://nandminc.gumroad.com/l/hvac-kit for $1.00 — the live page price as I publish this, read from the listing's own price metadata before and after writing. The page also displays SAR 3.75 as a local-currency estimate for a Saudi visitor; that is a display conversion, not a second price. For that you get the 17-file ZIP — the pricing guide, three estimate templates, five worked jobs, the startup ladder, the maintenance-agreement model, the business plan template, the scaling checklist, the SOP pack, the compliance checklist, both playbooks, the sales scripts, the 7-day launch plan and the self-checker — delivered through your Gumroad receipt link. Questions first? Email nandminc2020@gmail.com.
Licence in one line: personal use only for the purchaser — fill in and adapt the documents for your own HVAC business or client work, but no resale, no redistribution, and no bundling into another product or course.
FAQ
Do I need to know Python? No. The documents are plain Markdown and the two CSVs open in any spreadsheet. Python 3 is only for self_qos.py, and after the first integrity run you can ignore it entirely.
Can I use the formula in my own spreadsheet? Yes, and you should — just set your rounding to half-up. See leak two above. Excel's ROUND already is; Python's round() is not.
Will the numbers work in my market? The structure will; the figures will not. Replace the wage, the overhead lines and the per-truck costs with yours before you price a single job. The frozen examples are there to prove the formula, not to be copied.
Is the maintenance agreement model realistic? It is arithmetic, not a forecast. 190.00 per system per year against 142.72 of cost is the model; whether your customers sign at that price is a sales question the file cannot answer for you.
If you price HVAC work and you have a rounding or ordering story that bit you, I want to read it: nandminc2020@gmail.com. For more measured tool and template deep-dives, start at aistacklab2026.blogspot.com.
Comments
Post a Comment