The short version: a local business that needs software has three options, and two of them don't really work. The third one — pay a small amount every month for ongoing help — is the model almost nobody is offering yet, even though it's both cheaper for the business and much better for the person doing the work. Here's why, with the actual numbers.
Imagine Maya.
Maya runs a pottery studio out of a converted garage. Sundays she teaches classes. Tuesdays she fires kilns. The rest of the week she's fighting her spreadsheets.
Maya takes bookings — each class has ten spots and a waitlist. She also takes deposits, tracks inventory (clay, glazes, the tools she rents out), sends reminder emails, and occasionally refunds somebody whose kid had a soccer tournament. For the last three months, she's been losing a weekend every month cleaning up double-bookings and deposit confusion.
She's tried things. Calendly is too simple. Mindbody is $250 a month and built for chain studios. Squarespace has a booking plugin but it doesn't talk to her payment system. Her nephew mentioned “just use Notion.” She does not want to just use Notion.
What Maya actually wants is a booking-and-deposits app that fits her business — her class names, her waitlist rules, her cancellation policy. Something she can text her students about: “Hey, book here.”
She'd pay $400 a month for it, gladly. That's less than she pays her accountant.
Here's the problem: there's almost nobody to build it for her.
Her three options, and why two don't work
Option 1 — Hire an agency. She called one. They quoted $62,000 upfront and said they could start in October. It was March. Hard no for a studio that makes $18,000 in a good month.
Option 2 — Piece together SaaS tools. She's tried this for two years. Every tool handles about 60% of what she needs. Nothing handles the edges. She loses fees on mismatched integrations and spends hours a month moving data between systems by hand.
Option 3 — Hire a freelancer. She did this once. Nice person on Upwork, built her a working v1 in three months for $4,800, then vanished. The app works. Nobody can change it. Last month it stopped sending confirmations and she doesn't know why.
None of these match what she actually needs: someone who's around every month, who knows her system, who she can text with “hey can we add gift cards?” That “around every month” part is the whole thing.
The model nobody's offering
A monthly subscription for ongoing custom software. Pay $400/month. Get a working app. Email someone when it breaks. Ask for changes when the business changes. The person on the other end keeps showing up.
This is how Maya already pays for her accountant. For her phone, her internet, her payroll service, her credit card processor, her email marketing tool. Every adult bill she has is monthly. The only thing she's ever been asked to pay for in a lump, upfront, is — somehow — software built specifically for her business.
There's no structural reason for that. It's a historical accident. Agencies got big billing big projects, and the model stuck. Now it's the whole reason she can't get what she needs.
This is the gap. It's 33 million small businesses wide.
The math, which is strange and kind of beautiful
It turns out the monthly model isn't just better for the buyer — it's much better for the seller. Three numbers, from public research:
Monthly clients stay way longer. Agencies on monthly plans lose about 18% of clients per year. Agencies doing one-off project work lose 42% — more than two out of every five clients, gone every year, constantly replaced by new ones. The average monthly client sticks around 56 months. The average project client, 24. One is a business relationship; the other is a transaction.
The revenue compounds. If Maya pays $400/month for 56 months, that's $22,400. If she'd paid $4,800 once for a project and it fell apart two years later, that's $4,800. Same starting client. Completely different outcome.
Selling the agency one day works very differently. When an agency is acquired, the buyer doesn't care much about past projects — those already shipped. They care about future revenue. Agencies where 60%+ of revenue is monthly contracts sell for four to six times net profit. Agencies doing mostly project work sell for two to three times. The multiple is almost double. That's the exit math.
None of these numbers require the person doing the work to be a genius. They just require them to be on a monthly plan instead of chasing quotes.
Three people, three rungs
Maya's not the only shape of this client. Here are three realistic customer sizes:
Maya — pottery studio, $400/month. Smallest rung. You build her a booking + deposits + waitlist app. She pays you less than her accountant. You handle fixes and small changes every month.
Raj — dental practice, $2,000/month. His paper intake forms kill ten hours of front-desk time a week. You build a patient intake system. He pays you $2,000/month and calls it a bargain because he's saving $2,500/month in staff time.
Jenn — regional health insurer, $25,000/month. Their giant IT vendor quoted $2 million and 14 months to modernize a claims portal. She hires you instead. You ship in 90 days. At $25k/month you're a rounding error on her IT budget — and she never wants to deal with the big vendor again.
Maya's deal looks nothing like Jenn's on paper. The playbook is the same: monthly pay, ongoing relationship, ship changes as the business changes. What lifts you from Maya's rung to Jenn's rung isn't a new business model — it's what you know how to build.
Why right now, and why this stops being obvious soon
This model always made sense. It just wasn't possible until recently.
Twenty years ago, building a custom app for Maya would have taken a small team a year. Ten years ago, a senior engineer could do it solo in three months. Today, one person with the right training and AI handling the typing can do it in two weeks. The economics finally work for the $400/month price point — for the first time ever.
Here's the part nobody talks about: in roughly two years, most people are going to figure this out. Gartner thinks 80% of engineers will have to upskill on AI work by 2027. When everyone is doing this model, the operators who started early will be the ones small businesses already trust, already recommend, already call. That network forms narrowly and it's forming now.
So what do you do about it?
If you're reading this and thinking “I could do this,” you probably can. You need two things: the skill to build software that holds up (not just “it works on my laptop” — holds up when Maya's students are all hitting it at once), and a system to run the business without it turning into a second job.
The first part is what we teach. The second part is what we built.
Same playbook. Different rung. Start wherever you are.