Owners who run everything themselves tend to ask for "automation" as one big thing. It is never one big thing. It is fifteen small jobs, and two of them are eating most of the week.
Find those two first. Most of the time they do not need custom software at all.
Key Takeaways
- Track a normal week before deciding anything. Your memory overweights the annoying jobs.
- Score each job on hours, mistakes, and how often it repeats the same way.
- Automate jobs that repeat the same way. Leave judgment calls to people.
- Try an off-the-shelf tool before a build. Build when the tools force you to change how you work.
Step One: Write Down a Real Week
For five working days, keep a note open. Every time you do a task that is not the actual service you sell, write down what it was and roughly how long it took. Quoting. Confirming appointments. Chasing invoices. Copying details from an email into a spreadsheet. Answering "what time are you coming?"
At the end of the week you will have a list that surprises you. Owners usually guess that the job they hate most is the biggest. Often it is some small, boring task that happens thirty times.
Step Two: Score Each Job
Give each job a score from 1 to 3 on four questions:
| Question | 1 | 3 |
|---|---|---|
| How many hours a week? | under 1 | over 4 |
| How costly is a mistake? | annoying | loses money or a customer |
| Does it repeat the same way? | different every time | identical steps |
| Could someone else do it with a checklist? | no, needs your judgment | yes |
Add the scores. The highest totals are your candidates. A job that scores 3 on "repeats the same way" and 3 on "checklist" is a machine's job. A job that scores 1 on either should stay with a person, however many hours it takes, because automating judgment produces confident mistakes.
Step Three: Match the Job to the Cheapest Fix
Most high-scoring jobs fall into one of these, and the fix is usually cheaper than people expect:
- Appointment confirmations and reminders. Your booking or calendar tool probably already does this. Turn it on.
- Copying data between two apps. A connector tool like Zapier or Make. See Zapier vs custom software for where these stop.
- Chasing unpaid invoices. Your accounting software has automatic reminders.
- Answering the same five questions. A good FAQ page first. An AI assistant second, and only with a human handoff. Our guide to the cost to build an AI chatbot for a small business covers the risks.
- Producing quotes from your own price list. This is often where custom software starts, because every business prices differently.
Custom software makes sense when the job is central to how you make money and no tool fits it without forcing you to change your process. That usually means quoting, scheduling with your own rules, or a customer-facing view of their job.
A Concrete Version
A two-person residential cleaning company tracks a week. The list:
- Confirming next-day appointments by text: 3 hours a week spread over five days, a missed one means a no-show, identical steps. Score 11.
- Writing quotes from square footage and number of bathrooms: 5 hours, mistakes lose jobs, same formula every time. Score 12.
- Handling complaints: 2 hours, very costly, different every time. Score 7, and "repeats" and "checklist" are both 1. Stays human.
- Invoicing: 1 hour, already handled by their accounting tool. Just needs reminders switched on.
First move: switch on reminders in the booking tool they already pay for. Cost: zero. Saves 3 hours.
Second move: the quote. A simple form where the customer enters the size of the home and gets a price, using the owner's own formula. If the formula is a few lines, an AI builder can do it in a weekend. If it has exceptions (pets, first clean, recurring discounts) and has to create the booking too, that is a one-developer job in the $3,000 to $12,000 range our honest read uses.
The Honest Counterpoint
Automating a broken process makes it break faster. If your quotes are wrong because the formula itself is wrong, software will produce wrong quotes at scale and with more confidence. Fix the process by hand for a few weeks, then automate the version that works. And some jobs are worth keeping manual because they are where customers feel looked after. A personal call after a complaint is not waste.
Frequently Asked Questions
How much should I spend on my first automation?
Start with what you already pay for. Most owners have features switched off in tools they own. After that, spend in proportion to hours saved. A job that takes five hours a week is worth a few hundred dollars a month to fix, or a one-time build that pays back within a year.
Should I hire a virtual assistant instead?
For jobs that need judgment but not your judgment, often yes. An assistant plus a checklist is flexible in a way software is not. Automate the parts of their work that repeat identically.
What is the most common mistake?
Buying a big all-in-one platform to fix one job, then spending months adapting the business to the platform.
When do I need a developer?
When the job is core to how you earn, no tool fits it, and you have tried doing it the same way by hand long enough to write the rules down. Our post on how to turn a spreadsheet into an app is often the next step.
The Bottom Line
Track a real week, score the jobs, and start with the cheapest fix for the highest score. When a job turns out to need custom software, describe it in the honest read to get a cost range and a verdict. If the verdict is a build, our software factory works at a fixed price.
Roberto Espinoza is CEO of Ruzora, which builds custom software for business owners at a fixed price and places pre-vetted senior LATAM engineers with US teams. Get a free honest read on your idea.
