Custom web app vs SaaS, when does building actually beat buying?

TL;DR

Buy SaaS first. It's nearly always the right starting move, faster to launch, no maintenance burden, predictable monthly cost. Build custom only when one of three signals is loud: (1) the SaaS forces a workaround that costs money or customers every week, (2) per-seat fees are eating $300+/mo and growing, or (3) the workflow your business runs on doesn't exist as a SaaS at any price. Two of three is "build now"; one is "wait and watch."

The build-vs-buy decision gets pitched as a math problem (what does it cost to build vs what do we pay per month?) but the better lens is strategic. SaaS is a great default because most workflows are similar to most other workflows; the SaaS owners have a billion data points and have shaped a generic-good product. For a small business doing standard work, that's usually fine.

The signal that you've outgrown SaaS is friction at the edges. Your team is keeping a parallel spreadsheet because the SaaS report is wrong. The SaaS workflow forces "create a fake record so the integration works." Per-seat fees are a meaningful share of payroll. A workflow your business runs on (the way you quote, the way you book, the way you onboard) doesn't exist as a SaaS at any price. Each of these is a small thing; together they mean the SaaS is the wrong shape.

Custom web apps in 2026 cost less than they used to. Foundation models compress writing. Frameworks (Next.js, Remix, SvelteKit, the AI SDK) compress wiring. Deploy platforms (Vercel, Cloudflare, Railway, Fly) collapse infrastructure. A focused small-business custom app that would have been $80K in 2018 is $15K in 2026, with a smaller maintenance footprint.

The right pattern for most growing small businesses: SaaS for the boring layers (accounting, payroll, calendar), custom for the layer that's genuinely yours (intake, quoting, the way you serve customers, the way you run operations). Hybrid wins; pure-SaaS underwhelms at scale; pure-custom over-invests early.

Key facts

Custom small-business web app: typically $5,000–$30,000 build, $50–$300/month run.
SaaS equivalent (when one exists): typically $50–$500/seat/month.
Build payback period: 6–18 months for tools replacing $300+/month SaaS plus ~5 hours/week of human work.
Hybrid (buy infrastructure, build the layer that's yours) is the 2026 default for growing small businesses.

Common follow-ups

How do I know if my workflow is "yours" vs generic?

Ask three customers why they hired you. If "the way you handle quotes" or "your intake feels different" comes up, the workflow is part of why they paid. If they say "you were available" or "you had good reviews," it's incidental and SaaS will fit.

What if I migrate from SaaS to custom later?

Yes, and this is the right path for most. SaaS first to learn the workflow, custom build later when the ceiling is obvious. Building from day one without knowing the workflow tends to ship the wrong thing.

Sources

a16z: Build vs Buy

By Isaiah Grant, Founder, Rebuilt StudioUpdated Apr 30, 2026