# Should You Build a SaaS? The Honest Answer

> Source: [https://botensten.com/articles/should-you-build-a-saas](https://botensten.com/articles/should-you-build-a-saas) (canonical)
> Author: Botensten — Botensten, https://botensten.com
> Published: 2026-08-25

## TL;DR

Yes, build a SaaS if you can charge for a recurring problem and land a paying customer within about 90 days. No, if you cannot name real buyers, have no distribution, or need income this quarter. SaaS pays off through retention and compounding recurring revenue, not a launch spike. Validate demand with pre-sales before writing much code. Start with a narrow wedge for one type of buyer, ship a thin version fast, and expand only when churn is low and customers ask for more.

Build a SaaS only if you have a recurring problem people already pay to solve and you can reach your first paying customer within 90 days. The biggest killer of software startups is no market need, which [CB Insights' analysis of startup post-mortems](https://www.cbinsights.com/research/startup-failure-reasons-top/) ranks as the top reason for failure. If you cannot name ten people who will pay you this month, build a smaller tool or a service first.

## Should you build a SaaS, or something simpler?

Build a SaaS when the problem repeats monthly and buyers will pay to make it go away every month. Build something simpler when demand is unproven, because a service, a spreadsheet template, or a single script tests the same idea in days instead of months.

SaaS earns its keep through retention. One customer paying $50 a month for two years is worth far more than a one-time $200 sale, but only if they stay. That "if they stay" is the whole game, and it is why validation beats building.

Cheaper tests before you commit:
- Sell the outcome as a manual service first, then automate the parts customers pay for most.
- Offer a paid pre-order or annual plan before the product exists.
- Ship a no-code version (Airtable, a form, a Zap) and watch what breaks.

Eric Ries's [The Lean Startup](https://theleanstartup.com) calls this the minimum viable product: the smallest thing that produces real learning. The point is not to build less for its own sake. The point is to spend money only after a stranger has spent theirs.

## What does it actually cost to build a SaaS in 2026?

A solo founder can ship a first paying version of a SaaS for under $100 a month in tools plus their own time. The real cost is not servers. It is the months of focus you spend before revenue covers your rent.

| Path | Time to first version | Monthly tooling cost | Best for |
|------|----------------------|---------------------|----------|
| No-code (Bubble, Airtable, Zapier) | 1-2 weeks | $30-$100 | Testing demand fast |
| AI-assisted code (you + an AI pair) | 3-8 weeks | $50-$150 | Owning the product long term |
| Hire an agency | 3-6 months | $15,000+ upfront | Funded teams, not solos |

Hidden costs matter more than hosting. Support, refunds, failed payments, and churn all eat time. Budget for the boring parts: a payments provider, transactional email, error logging, and backups. These are cheap individually and painful when missing.

## How do you know if there's real demand?

You know there is demand when strangers pay before the product is finished. Interest, likes, and "I'd totally use that" are not demand. A credit-card charge is.

Run this validation sequence in order:
1. Write the exact one-sentence problem and who has it.
2. Find ten people with that problem and talk to them, not at them.
3. Ask what they use now and what they pay for it today.
4. Offer a paid pre-sale or pilot at a real price.
5. Build only the slice the paying pilots demanded.

Paul Graham's essay [Do Things That Don't Scale](https://paulgraham.com/ds.html) argues that early founders should recruit users by hand, one at a time. That manual work is not a detour from building a SaaS. It is the market research no survey can give you, and it tells you what to build first.

## What we learned shipping SaaS features every day

We build production software with AI every day, and the lesson that keeps repeating is that shipping is easy now, so choosing what to ship is the hard part. AI let us cut a feature from a two-week task to a two-day one, which sounds great until you ship the wrong feature twice as fast.

Our rule: no new feature ships without a named user asking for it. Early on we built a slick analytics dashboard nobody opened. We tore it out. The feature that actually moved retention was boring — clearer email receipts and a one-click way to fix a failed payment. It was ugly, it took an afternoon, and it kept customers who would have quietly canceled.

The trade-off we hit constantly: speed versus surface area. Every feature you ship is code you now maintain forever. We now delete more than we add, and we treat "can we say no to this?" as the default question. Owning your software, instead of renting a stack you cannot change, only pays off if you keep that surface small enough to actually own.

## Which businesses should NOT build a SaaS?

Do not build a SaaS if you need income this quarter or you have no way to reach buyers. SaaS revenue starts as a trickle and compounds slowly, so it starves a founder who needs cash now.

Skip SaaS, at least for now, if any of these are true:
- You have no audience, list, or channel to reach your first 50 customers.
- The problem is a one-time job, not a monthly one.
- You cannot commit 12 or more months before it pays a salary.
- You are excited by the tech, not the customer's problem.

The healthiest reason to build is that customers already pay you to solve this by hand and you are drowning. That is a business asking to become software. The riskiest reason is that SaaS sounds like passive income. It is not passive, and the recurring revenue only recurs if the product keeps working and you keep answering support.

## Should you build or buy the software your business runs on?

Build the software that is your edge, and buy the software that everyone else also uses. If a tool encodes how you win — your pricing, your workflow, your data — owning it means you can change it the day the market changes. If it is generic (email, accounting, calendars), renting is cheaper and smarter.

For most operators the answer is a mix. Rent the commodity layers, own the one system your business truly runs on. That is the renter-to-owner shift: you stop being at the mercy of a vendor's roadmap for the part of your stack that decides whether you win.

## Related reading

- [How Do I Build a SaaS? A Builder's Honest Playbook](/articles/how-to-build-a-saas)
- [What Do You Need to Build a SaaS? The Honest List](/articles/what-you-need-to-build-a-saas)
- [How Do I Launch a SaaS Product? The Operator Playbook](/articles/how-to-launch-a-saas-product)
- [Should You Discount Annual Subscriptions?](/articles/should-you-discount-annual-subscriptions)

## Frequently asked questions

**Should I build a SaaS?**

Build one only if you have a recurring problem buyers already pay to solve and you can land a paying customer within about 90 days. If you cannot name real buyers or need cash this quarter, start with a service or a smaller tool instead.

**How much does it cost to build a SaaS as a solo founder?**

You can ship a first paying version for under $100 a month in tools plus your own time. The larger cost is the months of focus before revenue covers your bills.

**How do I validate a SaaS idea before building it?**

Talk to ten people with the problem, learn what they pay for today, then offer a paid pre-sale or pilot. Build only the slice paying customers actually asked for.

**How long before a SaaS makes money?**

Plan for 12 or more months before it pays a salary. SaaS revenue starts small and compounds through retention rather than a one-time launch spike.

**Is SaaS really passive income?**

No. Recurring revenue only recurs if the product keeps working and you keep answering support, handling failed payments, and reducing churn.

**Should I use no-code or write real code for my SaaS?**

Use no-code to test demand fast in one to two weeks. Move to AI-assisted code when you want to own the product long term and control your core workflow.

**When should I build software instead of buying it?**

Build the software that is your competitive edge — your pricing, workflow, or data. Buy generic tools like email, accounting, and calendars that everyone else also uses.