Skip to content

Build DecisionsNovember 5, 202510 min read

By Usmani - Founder & Lead Engineer

Build vs buy software: the honest decision framework

SaaS subscriptions, spreadsheets or custom-built: each wins in different situations. Here's the framework we use with clients before anyone writes a line of code.

Every business runs on three kinds of software: the tools you rent, the spreadsheets you improvised, and, if you're lucky, the system you actually own. Build vs buy is really the question of which workflows deserve to graduate out of the second category.

This is the framework we walk clients through on scoping calls. It's deliberately boring, because the right answer usually is.

The three options, honestly

OptionCost shapeBest when
Buy (SaaS)$50–300/user/mo foreverThe workflow is generic, email, docs, payroll, video, accounting
Spreadsheet + duct tapeFree-ish, paid in hoursOne person owns it, volume is low, errors are cheap
Build (custom)Fixed cost upfront + small running costThe workflow is yours, recurring, and doesn't fit anyone's template

The five questions that decide it

  1. Is this workflow your differentiation? If it directly produces revenue or customer experience, owning it compounds. If it's payroll, rent it.
  2. Does any existing tool actually fit? Not 'kind of, with workarounds and three Zapier steps', fits. Test this seriously before building anything.
  3. How many hours per month does the manual version cost? Multiply by your loaded hourly rate. That's your custom-build budget ceiling for a ~2-year payback.
  4. Will the process still exist in 3 years? Building for a workflow you're about to abandon is the most expensive mistake in this category.
  5. Who owns it when the builder leaves? If the answer is 'the freelancer's laptop', you haven't bought software. You've rented a person.

The math, on a real example

A property management company doing tenant requests over WhatsApp: ~15 staff-hours a week of logging, chasing and updating. At $15/hr loaded that's ~$11,700/year. Every year, forever, and it scales badly with headcount.

A custom tenant portal + dispatch system: ~$4–5K once, ~$200/month to run. Payback inside six months, and the system doesn't forget, quit or lose the message thread.

When build loses. Say it plainly

  • The workflow is genuinely generic. Nobody wins building their own email or payroll.
  • You're pre-revenue and don't know what the process even is yet. Start with a spreadsheet, learn, then build.
  • There's no budget for maintenance. Unmaintained custom software dies slower than SaaS but still dies.
  • A decent SaaS gets you 85% of the way. That last 15% is usually not worth five figures.

The hybrid answer most companies actually need

Almost nobody should build everything. The pattern that works: buy the commodity layers (email, accounting, storage, auth), build the workflow layer that's actually yours, and integrate them. That's exactly what we mean by 'one system from intake to invoice', your process, riding on commodity infrastructure you didn't have to build.

See how a custom build gets scoped

Custom software development

Frequently asked questions

When is custom software worth it?

When the workflow is core to how you make money AND off-the-shelf tools force you to work their way. Custom pays back when it removes recurring manual hours or unlocks revenue the tools couldn't reach.

Isn't custom software much more expensive than SaaS?

Upfront yes: $1.5K+ vs $50/user/month. Over 3 years the math often flips: per-seat SaaS for a team plus the manual work it doesn't automate frequently costs more than one fixed-price build.

What if my needs change after it's built?

That's the advantage of owning it: changes are a quote, not a feature request sent to a vendor's roadmap. A good build is modular. You extend the parts, not replace the whole.

Should we ever build something a SaaS already does?

Almost never for commodity functions (email, payroll, video calls). Build the 20% that's your actual differentiation; integrate the rest.

Next steps

All insights