All SMB· Cost & ROI

Custom Software Quote vs Operator-Built App: The $200k vs $5k Breakdown

Agency quotes for custom business software often start at six figures. A $20M Spanish construction company built its own system and now runs the operation on about $5k/year. This is how the math works in 2026.

A $20M Spanish construction company owner got the same quote 60% of SMB owners get when they shop for custom software: $200,000.

The agency's pitch was reasonable. Two senior developers. Six months. iOS, Android, web. Project manager. Infrastructure setup. App Store submission. A maintenance retainer of $5,000/month after launch.

His wife is a project manager at a Big 4 consulting firm. She looked at the proposal and said the quiet part out loud: "In Deloitte, we can build something very, very good. But this is not for construction. We are not people in relation with construction. If you ask them, of course they can build it. But for how much?"

He didn't sign the agency contract. He built it himself in Rork over a few months of evenings. Today his company, $20M in annual revenue, 50 direct employees, 300 contractors, runs end-to-end on a system he made. He pays $200/month for Rork Max plus another $200 in cloud infrastructure. About $5,000 a year. Calls it a bargain.

This is the math that broke in 2026.

The Three Quotes Every SMB Owner Gets

When you ask the market "how do I get custom business software," you receive these three quotes:

Quote 1: The Agency Build

  • $80k to $200k upfront.
  • Plus $10k to $30k for a "discovery phase" before they commit to a number.
  • Plus $3k to $8k/month maintenance retainer.
  • Plus scope creep at every conversation past month two.

Total 3-year cost: $180k to $500k. You own the relationship, not the team.

This is the quote that triggers the search that lands you on guides like this one.

Quote 2: The Per-Seat SaaS Stack

  • CRM with mobile app: $80/seat/month × 25 field seats = $24k/year
  • Field service tool: $100/seat/mo × 15 = $18k/year
  • E-sign tool: $25/user/mo × 10 = $3k/year
  • Scheduling: $60/seat/mo × 10 = $7.2k/year
  • Document storage: $15/user/mo × 50 = $9k/year

Total: $50k to $80k/year. Growing every year as vendors raise prices. Your field team has five logins. Nothing talks to each other natively.

Quote 3: Hire a Developer

  • Senior mobile developer: $120k to $180k/year all-in (US), less elsewhere.
  • They will need a teammate within 6 months.
  • Plus CTO advisor: $30k to $60k/year.
  • Plus tooling, equipment, infrastructure: $5k to $10k/year.

Total 3-year cost: $500k to $900k. You're now running a tech team while still trying to run your construction company.

The Fourth Option: Build It Yourself

The one this guide exists to make visible:

Quote 4: The Operator Build

  • Rork Max subscription: $200/month = $2,400/year (or $20/month without native iOS Swift).
  • Supabase Pro: $25/month = $300/year.
  • Cloud storage (R2 or Supabase): $50/month = $600/year.
  • OpenAI API for AI features: $50/month per 20 techs = $600/year.
  • Apple Developer Program: $99/year.
  • Google Play Developer: $25 one-time.

Total: ~$5,000/year all-in. ~$15,000 over 3 years.

What you get:

  • Real native iOS and Android apps, on the App Store or via internal distribution.
  • Real web app on the same project.
  • Your Supabase database, your code, your accounts. Walk away from Rork tomorrow and your apps keep running.
  • No agency retainer. No employee payroll. No CTO to manage.
Model3-year costReal nativeYou own itManage devs?
Agency$180–500kNo
SaaS stack$150–250k✅ TheirsNo
In-house team$500–900kYes
Operator build~$15kNo

Why the Quote Was Always Going to Be $200k

Software agencies don't price arbitrarily. They have real costs (developer salaries, project managers, infrastructure, overhead, profit margins). A real custom iOS + Android + web build with 6 months of development time genuinely costs them around $100k internally, and they charge $200k to make a healthy margin.

The question isn't whether the agency is dishonest. The question is whether you need an agency.

In 2023, you did. AI tools couldn't produce real production-grade business apps. You needed humans typing.

In 2026, you don't. Tools like Rork compress what used to be six months of agency work into ~30 days of evenings for a non-technical operator who knows their business. The agency's value proposition shrinks.

The Three Things That Changed

1. AI Coding Got Good Enough for Real Mobile

In 2023, asking AI to "build me a native mobile app with offline sync, push notifications, and a camera flow" produced a toy. In 2026, the same request through a purpose-built builder produces a real native app running real business workflows.

2. Expo Standardized Cross-Platform Native

You used to choose: build iOS and Android separately (double the cost) or build a web app (and lose the real-app experience). Expo (which Rork uses) compiles to real native iOS, real native Android, and a real web app from one codebase. Same stack Discord, MLB, and Coinbase ship their mobile apps on.

3. App Store Distribution for Business Use Got Simple

Apple Business Manager Custom App distribution and Google Play Managed Distribution let you ship internal-use apps to your team without a public listing, without consumer App Review hassles, without marketing. Free for any business.

What the $20M Construction Operator Actually Got for ~$5k/year

The system that replaced his $200k agency quote:

  • Mobile field crew app (iOS + Android): GPS check-in, voice notes, photos, AI-generated site reports, offline sync.
  • Web app for the office (same Rork project): client pipeline kanban, contracts, invoicing, real-time map of where field crews are, cash flow.
  • Push notifications that wake him up when a new lead lands.
  • One-button contract generation with e-signature integration.
  • Three platforms live, one codebase, one Supabase backend.

His finance director told him a month after rollout: "Mickey, I'm doing all my work very fast."

His agency quote would have promised the same and shipped a 70% solution by month 9. He shipped a 100% solution by month 3, for a fraction of the agency price.

The Honest Tradeoff

The DIY model has a real cost the agency model doesn't: your time in the first 30 to 60 days. You'll spend evenings learning Supabase row-level security policies. You'll get stuck on Apple's bundle identifier dance. You'll rebuild a screen twice because you didn't think about iOS push notification permission prompts.

For some operators, that's a dealbreaker. They'd rather write the $200k check and have their weekends back. The agency model exists for them.

For most operators we've talked to, the answer is: "I'd happily spend 60 evenings to never pay another agency retainer, and to own the thing forever."

Which math fits your life is yours to decide.

What to Do This Week

If you have an open $200k agency quote on your desk:

  1. Don't sign yet. Open Rork.
  2. Describe the most painful workflow in your business. Use plan mode.
  3. Ship a single screen by Friday. Install it on your phone via TestFlight.
  4. By the end of next month, you'll know whether the DIY path works for you.

If it does, you save ~$195,000 in year one (and the option value of having shipped it yourself).

If it doesn't, the agency is still there. You haven't lost anything except a couple of weekends finding out.

See also:

Frequently asked questions

Is a $200,000 agency quote unreasonable in 2026?+
It's reasonable for what the agency offers (two senior developers for 6 months, project management, infrastructure setup, App Store submission, maintenance retainer). It's unreasonable as the only path forward, because in 2026 a non-technical owner can produce equivalent custom software for ~$5k/year using AI builders like Rork. The price is no longer correlated with what's actually needed to build the software.
Can a ~$5k/year DIY build really match a $200k agency build in quality?+
For most SMB internal-use apps, functionally yes. The bottleneck for SMB software isn't engineering complexity, it's knowing the business well enough to design the right workflow. Operators know that better than any agency. The agency build wins for complex consumer apps with novel UX, real-time multiplayer, or heavy custom native modules. Those aren't most SMB internal-use needs.
What's the catch?+
Your time. 30 to 60 days of evenings to ship a v1. You'll learn Supabase, App Store submission, push notification flows. Most operators say it's worth the trade, never paying an agency retainer beats spending weekends.
What if I need ongoing development and updates?+
Once the v1 is shipped, ongoing changes are usually fast (hours, not weeks). You describe the change to the AI builder, it makes it, you ship. Agencies bill for every change order. SMB owners with their own AI builder don't.
What about complexity beyond CRUD apps?+
Real native mobile apps with AI workflows, offline sync, push notifications, e-signature integration, multi-tenant access, are all within reach via Rork plus Supabase plus standard APIs. Where DIY breaks down: real-time multiplayer games, complex AR/VR, native video codecs, heavily regulated fintech. For 95% of SMB needs, DIY now matches agency.
How do I evaluate if I should DIY or go to an agency?+
Three questions. (1) Does your business have a unique workflow no SaaS vendor matches? (2) Do you have 30 to 60 days of evenings to invest in v1? (3) Are you willing to learn basics like database schemas and App Store submission? Three yeses = DIY. Three nos = agency. Mixed = hybrid (DIY core, contractor for niche pieces).

Related guides