How the work happens now
Every step, system, and handoff. Who owns what, how many times a year it runs, and where it breaks.
One paid piece of work
Find out what one expensive job is really costing you.
You know which job hurts. What you don't know is what it costs you in a year, or whether you're already paying for the fix. Name it and in two weeks I put a dollar figure on it, work out how much of it the software you already pay for can handle, and mark what has to stay with your people. Then I tell you whether to fix it, buy something, build something, wait, or leave it alone.
What you get
Enough to make the call, and to defend it.
Everything stays on one job. You end up knowing what it costs you and what to do about it.
What it costs you, in dollars
A conservative number for the hours and money at stake, the assumptions behind it, and what has to be true for it to pay off.
What you already own
The tools you pay for today, where setting them up right is enough, and where a gap remains.
How you'd test it safely
A small test with a written pass/fail line. Rules about which data gets touched, who reviews the output, and what happens when it's wrong.
The recommendation, with a plan
Fix what you own, buy, build, wait, or leave it alone. Then the order of work, who you need, and a budget range.
What the arithmetic looks like
I have not done one of these for your trade yet. Here is how I get to a number.
These numbers are made up, and I have used a plumbing company because the arithmetic is easy to follow. Yours will look different. What is real is the method: where every number comes from, and which one I argue with hardest.
- 01
Count the job
Say intake runs 40 times a week. Not an estimate from a meeting. We count it, from the phone system, the scheduler, or a week of tally marks by the person who does it.
- 02
Time the waste, not the work
Say 20 minutes a job goes on chasing the gate code, the model number, and what is actually broken, then keying it in twice. That is 20 minutes that would not exist if the information arrived complete. Doing the job properly is not waste.
- 03
Turn it into hours
40 jobs a week at 20 minutes is about 13 hours a week. Across 50 weeks that is roughly 700 hours a year, or about $23,000 of office time at $35 an hour loaded.
- 04
Then the money you never see
Say 3 jobs a week miss same-day scheduling because the details land too late, and the customer calls somebody else. At an average job of $450 that is about $68,000 a year of work you never booked. Call the job $90,000 a year.
- 05
Argue with the biggest number
That $68,000 is the one I push back on hardest, because it is the easiest to inflate and it is most of the total. If you would not actually have had the crew free, it is not real, and I take it out. A number you do not believe is worth nothing to you.
- 06
Then check you cannot already fix it free
Before anyone builds anything, I find out whether the scheduling software you already pay for takes a proper intake form. If it does, the answer is a settings change and a morning of training, and you should not pay me or anybody else to build something.
How it runs
Start with the job, not the technology.
I look at the software you already pay for, and at how the work is done now, before I go anywhere near building something. If the answer is a setting nobody turned on, a call to your vendor, better training, or just doing a step in a different order, that is what I will tell you.
- 01
Pick the job
We agree on which job, where it starts and stops, who runs it, who signs off, and what makes a fix worth paying for.
- 02
Watch it run
I map how the work happens today: every step, system, and handoff, how many times a year it runs, where it jams, and what your people do twice.
- 03
Test it
First I check whether software you already own can do it. Only what's left gets tested, using fake or masked data unless you approve real records in writing.
- 04
Give you the answer
You get the numbers, the risks, and the options, so you can go ahead, change course, wait, or drop it.
Existing software first
Never build what you already own.
Your portals, inboxes, CRM, document tools, scheduling software, and whatever automation came with them probably already cover part of this job. I check what you pay for, how it is set up, and why it is not doing what you wanted.
Turning on something you own, getting your vendor to do it, training your people, or changing the order of the work can all be the right answer. Nothing I recommend has to involve AI.
Decision criteria
What you should know before you spend anything.
The exact measures depend on the job. I write down the numbers and the assumptions behind them. I do not hand you savings claims I cannot support.
- How many times a year the job runs and what it costs you
- What your current software can already do, and how much of it you actually use
- How long it takes, how often it gets done twice, how many odd cases come up, and what you pay in overtime or outside help
- What data anything would touch, how it is kept, and who has to check the work
- Who and what a fix would need, and who signs off on it
- What you would need to see before letting it loose on real work
Client participation
What I need from you.
- Somebody who can say yes, and somebody who actually runs the job day to day. In a small company that is often the same person.
- One visit to you and up to eight conversations with your people, unless we agree otherwise in writing
- Time with whoever does the work, whoever handles the money, and whoever looks after your systems
- Access to how the job runs today, what systems it touches, and whatever numbers you already have
- Your written go-ahead before I touch any real customer or client records, and somewhere safe to do it
- Answers within a day or two when I check assumptions with you, so the clock does not stretch
- The people named, the paperwork done, and your existing numbers to hand before the two week clock starts
Outside the Blueprint
Work that requires a separate decision.
- Building it, running it afterwards, or any promise that it launches
- Professional advice or autonomous decisions about tax, law, insurance, housing, safety, eligibility, pricing, claims, payments, or customer outcomes
- Moving you off your main system, cleaning up all your data, or replacing the software your business runs on
- Software licences, travel, outside legal or security opinions, or third-party audits
- Any certification, or any promise about savings, revenue, accuracy, or whether your people will use it
If it goes wrong
What I put at risk.
Compical is new and I am taking on my first clients now. Being early costs you something, because I cannot point you at ten companies like yours that I have already done this for. It also gets you a founding price, and the commitments below.
One price, not a day rate
You get one number for the whole job before anything starts. If it takes me longer than I planned, that is mine to absorb, not yours.
Either of us can stop after week one
If the first week shows there is no job here worth the money, I say so. You can call it off then too, for any reason. Either way you pay for that one week and anything already paid for the rest comes back.
You decide whether I answered it
Before I start we write down, in your words, what the final recommendation has to answer. You are the one who says whether it did. If you say it did not, you do not pay for it, and if you have already paid, I send it back.
Nobody pays me except you
No referral fees, no reseller margin, no commission from a vendor. If the answer is that you already own the fix, that is the answer you get.
The work about your business is yours: your numbers, your process, the recommendation. Take it to another firm if you want. I keep the general know-how and I may build software from patterns I see across businesses, but nothing of yours goes into it.
Start here
Which job is costing you the most?
The call is free and takes 30 minutes. I'll ask what's going wrong, what software you run, who makes the call, and when you'd want it fixed. If I'm not the right person, I'll say so.
Book a 30-minute call