Key takeaways
- Four things size an automation build: the systems it touches, its triggers, its approval steps and the data cleanup. Data cleanup is the line owners never budget.
- You get a set price for the initial build, quoted up front, then a monthly plan for only the services the system uses: hosting, texting, support, updates. A system that texts customers all day carries a different plan than one that pings one phone.
- The app-stack alternative bills every month: QuickBooks lists at $65, Jobber at $49 for one user plus $29 per extra user, route software at $50 to $199, and Zapier climbs with every task.
- Price the problem before the build: hours a week times what your hour is worth, plus what mistakes cost. Under an hour a week with no error cost, do not automate it yet.
- Automation only pays in a business that already has customers: 14 of my own software lines are automated end to end and have $0 in lifetime revenue.
The short answer: the cost to automate a small business depends on four things
There is no honest sticker price for "automation," because the word covers a text alert and a whole back office. What there is: a set price for the initial build, quoted up front once the job is scoped, and a monthly plan for the services the system runs on. We build every system around the business in front of us, so the scope sets the number.
Search this question and you will see published ranges from $1,500 to $40,000. Three things cause the spread.
- The word covers too much. "Automation" means a Zap that texts you when a form comes in, and it means wiring a CRM to an accounting system to a dispatch board.
- The pricing models hide the real number. Some agencies quote setup plus a retainer that never says what it pays for. Platforms price per seat or per task, so the bill grows with your crew and your leads. Developers price by the hour, so the number arrives with the invoice.
- Nobody says what is included. Carrier registration, list cleanup, the month of fixes after launch: each is real money, inside or outside the quote.
When we quote, the build price is set before work starts and includes the fixes after launch, and the monthly plan names every service it covers. The rest of this post is what moves each half.
What makes a build bigger or smaller
Most jobs fall into one of three sizes, and four things decide where inside a size a job lands.
- Smallest: a single alert or notification. My form-to-phone lead alert is this shape: a form submits, one small function pushes the name and phone number to my phone with a tap-to-call button, and an email goes out as the record. Texting the lead instead of pushing it adds a carrier registration, and that alone makes it bigger.
- Middle: a multi-step workflow. My outreach engine lives here: templates for five trades, a 25-a-day cap, random spacing between sends, suppression and bounce lists, an 8 a.m. weekday schedule and a tally pushed to my phone.
- Largest: a small app with its own database. My snow route app is a client list, storm records, a route in road order and a one-stop-at-a-time card. My bookkeeping app adds statement import, categorizing rules and a package for my CPA. Past that you are buying a platform, which gets a written scope of its own.
Systems touched
Every system is a login, an API, rules about what it accepts and a way it fails. My form alert touches two; the outreach engine touches four.
Triggers
A form submit is one trigger. A schedule is a second, with its own failure mode: the first version of my outreach engine did nothing when the Mac was asleep at 8 a.m. Now missed runs fire on wake.
Approval steps
A human in the loop is the most expensive line in any workflow. My outreach engine will not send live until someone types "send" at a prompt, and its default is a dry run. If someone has to approve a quote before it goes out, say so up front; it is often the difference between the smallest size and the middle one.
Data cleanup
This is the line owners never budget. When I built the outreach engine, the Minnesota lead list had 154 deliverable emails. The nationwide list had thousands of rows and was 98 percent Facebook pages with no email at all. The time went into the list, not the code. If your customer list lives in three places, the cleanup is part of the build, and it goes into the quote before work starts.
What the monthly plan covers, and why it varies with the services you use
The build gets the system working. The monthly plan keeps it working, and it covers only the services your system actually uses.
- Hosting. Somewhere for the functions, the pages and the database to live. A form alert needs very little; an app the whole crew opens on a storm morning needs more.
- Texting. If the system texts customers, the carriers bill for it. Twilio lists under two cents a message and about $1.15 a month for a number, plus a carrier registration before anything delivers. A system that texts a hundred customers a day carries a different plan than one that pushes an alert to one phone.
- AI usage. Categorizing bank lines or drafting a reply is billed by the model provider per use: pennies per statement in my bookkeeping app, more for anything that listens to phone calls.
- Support and updates. The morning a vendor changes an API, a carrier changes a rule, or you want a new field on the form. Someone has to own that, and on the plan it is our development team.
That is why I will not put one monthly number on a page; it would be wrong for most readers. What you get instead is a plan that lists each service and why your system needs it, so you can see what every line is for and leave out what you do not use.
It is priced to compete with the subscription apps you would otherwise stack.
Price the problem first: hours times loaded rate plus error cost
Before anyone talks about a build, work out what the task costs you now. The formula I run before I scope anything:
Annual cost of the task = hours per week × loaded hourly value × 50, plus the monthly cost of mistakes × 12.
Loaded hourly value is not your wage. It is what the hour produces when you spend it on the work you sell: close to your billing rate, or last year's take divided by 2,000.
A typical case: 3 hours a week, an owner whose hour is worth $75, and $100 a month in mistakes. That is $11,250 plus $1,200, or $12,450 a year, about $1,037 a month, and roughly $37,000 over three years if nothing changes. That is the number any quote, or any subscription, has to make sense against.
The error cost is the part owners skip. Before the bookkeeping app my books lived in a spreadsheet, and a running balance got pasted in as an amount: a fast-food lunch showed up as an $18,818 expense. The parser I built later caught it, because it reconciles every statement section to the penny. Use a conservative monthly number and know that it is low.
Now my weakest case on hours alone, honestly. The bookkeeping app took my books from about 3 hours a month to about 12 minutes a week, only about 26 hours a year, around $2,200 at my rate. On hours alone I would not have built it.
I built it because of the $18,818 lunch, and because the other option on the table was a QuickBooks seat at $65 a month (Intuit's list price) for software made for everyone's books instead of mine. Run the calculator below with your own numbers before you believe mine.
When automation isn't worth it: the $0-revenue lesson
In September 2026 I ran a revenue analysis across everything I own: 1,229 transactions over 18 months. Every operating dollar, 38 deposits, came from Revisions Unlimited, all of it word of mouth. On the other side of the ledger sat 14 software product lines, QuoteShield among them, with $0.00 in lifetime revenue.
Every one of those 14 is automated beautifully. The template store has 58 products wired to Stripe checkout. Another of the 14 has a full Stripe Checkout flow with a webhook behind it, ready to fire the second someone pays. Nobody has bought from either. The automation works perfectly and automates nothing, because there was never a customer on the other end.
That is the lesson every "automation cost" article leaves out: automation multiplies a business that already has customers. It cannot create one.
Before you spend a dollar, check these five. If any one is true, wait.
- You do not have customers yet. Get the first ten before you automate.
- The task is under an hour a week and a mistake costs nothing.
- The process changes every month. Run it by hand until it settles.
- The data is garbage. A perfect sending engine with no email addresses sends nothing.
- Nobody owns the exceptions. My bookkeeping rules place 88 percent of lines on their own; the other 12 percent still need a person once a week. If no one will be that person, the pile just moves.
Three systems in this post earn their keep because Revisions Unlimited has customers who fill out forms, need plowing and generate receipts. The outreach engine only earns its keep if it produces clients for Style Marking, and I count it that way: a bet on new work, not hours saved.
The app-stack alternative: Zapier, QuickBooks, Jobber and per-seat app builders
I sell builds, so read this knowing that.
Small jobs: a Zap is often enough
If your form tool can already text you, do that and keep your money. If not, Zapier's entry paid plan is about $20 to $30 a month depending on billing (zapier.com/pricing), and a single alert fits inside it.
The catch I hit on my own alert: a hidden spam-trap field was being filled by the browser's saved-address autofill and silently throwing away real bookings. A Zap will not tell you that. A build with a log will.
Workflows: no-code until the rules show up
A multi-step Zap handles "form in, row added, email sent." It gets brittle where my outreach engine lives: a daily cap, random spacing, suppression, a bounce log and a typed approval. Each rule is more steps, and per-task billing charges for every one on every run. Past five or so rules, a workflow built for how you actually work is easier to change, and the task bill is the number it has to compete with.
Apps: per seat, every month
Glide, Softr and Airtable price per user per month, and so do the trade tools: Jobber is $49 a month for one user on Core and $139 on Connect, plus $29 per extra user; route software runs $50 to $199 a month; QuickBooks lists at $65 a month for the plan I would have needed. For a crew, the seat count is the cost, and it grows with every hire.
My snow route app does road-time ordering, GPS re-planning and a finish-by estimate checked against a 12-hour promise, none of which a table-with-a-form can do. Generic apps stitched together make you work their way; one system built around how your business works does the job you actually have. The per-truck math and the QuickBooks version are worked out in their own posts.
How to get a set build price in one conversation
You get one build price, up front, and it is the build price you pay.
A set price forces the scope to be written before money changes hands. When the scope comes first, the four things above are on paper, and anything not on the paper is a change we price separately, in writing, before it is built.
I also do not give design or build work away to win a job. A mockup is the design, and a shop that hands it out earns it back somewhere in a price you cannot see. The free thing I offer is the quote: you describe the task, we come back with a written scope, a set build price and the services the monthly plan would cover, and you can take that scope to anyone.
Our development team works hard for every client, and you talk directly with the people building your system. We work remotely with owners anywhere, and in Minnesota the team can come out in person.
To get a number in one conversation, send three things through the contact page:
- The task in one sentence, with its trigger. "When a form comes in, text me." "Every Monday, remind the open invoices."
- The systems it touches, by name. Your form tool, your calendar, your accounting, your phone.
- Hours per week it costs you now, and what one mistake costs. Rough is fine.
With those three we can tell you on the first reply how big the build is and which services it needs, and send the set build price on the second, priced to compete with the apps you would otherwise stack. If the math says wait, we will say so. The lead alert, missed-call text-back and the key-man post go deeper on three of the builds above.
Frequently asked questions
It depends on what the system has to do: how many tools it connects, what triggers it, where a person approves, and how clean the data is. You get a set build price, quoted up front after a short scoping conversation, then a monthly plan that covers only the services the system uses, such as hosting, texting, support and updates. It is priced to compete with the subscription apps you would otherwise stack. For a real number, ask for a free quote through the contact page.
Yes, if two things are true: you already have customers, and the task you want to automate costs you three or more hours a week or produces mistakes that cost real money. If either is false, wait. I have 14 software lines that are automated end to end and have never earned a dollar, because there was no customer on the other end.
Only the services your system runs on: hosting, texting if it messages customers, email sending, any AI usage, and support and updates when a vendor changes something or you want a change. That is why plans vary: a system that texts customers all day uses more than one that pushes an alert to the owner's phone.
The apps bill every month, per seat or per task: QuickBooks lists at $65 a month, Jobber at $49 for one user plus $29 per extra user, and Zapier climbs with your task count. A system built for your business has a set build price and a monthly plan for the services it uses, priced to compete with that stack. List what you pay now and ask for a quote to put beside it.
Start with what the task costs now: hours per week times your loaded hourly value times 50, plus the monthly cost of mistakes times 12. For a three-year view, take three years of that, subtract the build price and three years of the monthly plan, and divide by what you spent. Use your billing rate, or last year's take divided by 2,000, as the hourly value.
In my experience there are five causes: there were no customers yet, so there was nothing to automate; the task was too small to matter; the process kept changing; the data feeding it was dirty; and nobody owned the exceptions the automation could not handle. The code is rarely the reason.
Want a real number for the task you hate most?
Tell me the task, the systems it touches and the hours it costs you each week. You'll get a written scope with a set build price, quoted up front, and a monthly plan for only the services it uses, priced to compete with the apps you'd otherwise stack. Our development team builds every system around how your business works, and you talk directly with the people building yours.
Founder, Style Marking · owner, Revisions Unlimited. Jacob builds custom websites, apps and operations systems for small-business owners — and runs a remodeling company on the same tools he builds for clients. Every post here comes from systems he has actually shipped.