What it costs to run a small SaaS
The infrastructure is not what costs you. The subscriptions you signed up for because a blog post said you needed them are what costs you.
The short version
- Under $50/month covers a real product with paying customers on it.
- The free tiers are genuinely generous. Most products never outgrow several of them.
- Payment processing is your largest variable cost, and the one you cannot avoid.
- Tools, not infrastructure, are what inflate the bill. Audit them quarterly.
- Price so 30 customers covers everything, including your own time.
What does it actually cost at the start?
Close to nothing. A deployed product with auth, a database, transactional email, error tracking and a domain runs on free tiers plus about $15 a year for the domain. The first real bill usually arrives when you take your first payment, not when you get your first user.
| Layer | Typical free tier | When you start paying |
|---|---|---|
| Hosting | Enough for a real app | $5–$20/mo, once traffic is steady |
| Database | 0.5–1GB managed Postgres | $0–$25/mo at real data volume |
| Auth | 5,000–50,000 monthly users free | Much later, if ever |
| Transactional email | ~100/day, or 3,000/mo | $0–$20/mo |
| File storage | 1–5GB | Pennies per GB after |
| Error tracking | 5,000 events/mo | $0–$26/mo |
| Analytics | Free, or ~$9/mo privacy-first | Immediately, if you want good ones |
| Domain | None | $10–$15/year |
The number that surprises people is auth. Managed authentication is the layer founders assume will be expensive and it is the one with the most generous free allowances: most small SaaS products never leave the free tier at all.
Do not self-host to save money at this stage. A $20/month managed database that never wakes you up is cheaper than a $5 VPS you administer yourself, once you price your evenings at anything above zero.
What does the bill look like by stage?
| Stage | Monthly | What changed |
|---|---|---|
| Building, pre-launch | $0–$5 | Free tiers plus a domain |
| Launched, first customers | $20–$50 | Hosting off free tier, a paid email plan |
| $1k MRR | $50–$150 | Database grows, analytics, a support inbox |
| $5k MRR | $150–$400 | Real infrastructure, backups, a few paid tools |
| $10k MRR | $300–$800 | Still tiny relative to revenue |
The important shape here is that costs rise far more slowly than revenue. At $10k MRR a bill of $800 is 8% of revenue, and gross margins of 80 to 90% are exactly why software is worth building alone. Compare that to any business that ships something physical.
The second thing worth noticing: the jump from $0 to $50 happens at launch, and then very little happens for a long time. Founders brace for a scaling bill that mostly does not arrive at this size.
Which costs scale with users, and which do not?
Payment fees and email volume scale directly with your business. Hosting and database scale in steps, not smoothly: you sit on a plan for months and then move up. Tool subscriptions scale with your own discipline, and they are usually the largest avoidable line on the bill.
- Payment processing: scales exactly. Around 2.9% plus 30 cents per transaction, and there is no free tier. On a $29/month subscription that is roughly $1.14, or about 4% of revenue.
- Email: scales with sends. Cheap, but watch it: onboarding sequences multiply per signup, not per customer.
- Hosting and database: step functions. You outgrow a tier suddenly, not gradually.
- Storage and bandwidth: scales, but slowly. Only a real line if users upload files.
- Tools: scale with how many you bought. Fixed, recurring, and the easiest to cut.
The 30-cent component of the payment fee is worth a moment. On a $9/month plan it is 3.3% on its own, so low-priced monthly plans lose a disproportionate share to fixed fees: one of several arguments for pricing above the level where processing eats the margin, and for offering an annual option.
Where does the money actually leak?
Into tool subscriptions bought during a burst of optimism. A CRM, a project manager, a scheduling link, a form builder, a design tool, an email marketing platform, a status page: each $10 to $30, none individually alarming, together larger than your entire infrastructure bill.
- List every recurring charge from your card statement, not from memory. Memory misses about a third of them.
- Mark the ones you used this month. Actually used, not intended to use.
- Cancel anything unused for 60 days. You can always resubscribe; almost nobody does.
- Check for a cheaper tier on what remains. Most tools have one you skipped past at signup.
- Do this quarterly. It takes twenty minutes and it reliably finds something.
The tool that is genuinely worth paying for early is whatever removes a task you would otherwise do badly at 11pm: error tracking, automated backups, dunning for failed payments. Tools that organise work you could organise in a text file are the ones to cut.
What about the costs that are not money?
Your time is the real budget, and it is the one nobody puts on the spreadsheet. Ten hours a month of support, maintenance and firefighting is the actual price of running a small SaaS, and unlike the hosting bill, it does not have a free tier.
Three things reliably consume it, and all three are worth spending money to avoid:
- Support. Falls fastest when onboarding improves. Most tickets are the same three questions.
- Failed payments. Automatic retries and a dunning email sequence take an afternoon to set up and recover revenue you were silently losing.
- Anything you host yourself. Backups, upgrades, certificates, outages. This is the hidden cost in every “I saved money by self-hosting” story.
How many customers cover it?
Fewer than founders expect. At $29 a month with a $150 monthly bill, six customers cover the infrastructure. The number that matters is not break-even on the bill. It is how many customers it takes to be worth your evenings, and that is a much larger number.
| Price | Covers a $150 bill | For $3,000/month |
|---|---|---|
| $9 | 18 customers | 348 customers |
| $29 | 6 | 111 |
| $49 | 4 | 66 |
| $99 | 2 | 33 |
The right-hand column is the honest one, and it is the argument this whole article exists to support. Infrastructure is not what decides whether a solo SaaS works: 348 customers and 33 customers are wildly different businesses, and the difference is entirely a pricing decision made before you wrote any code.
So the operating cost question resolves quickly: keep the stack boring and mostly on free tiers, audit your subscriptions quarterly, and then spend your attention on the number that actually moves (what you charge, and how many people reach the point of finding it worth paying for).
Frequently asked questions
Can I really run a SaaS for free?
Before launch, essentially yes: free tiers plus a domain. After launch you will pay something, but $20–$50 a month covers a real product with paying customers on it for a long time.
What is the biggest cost at the start?
Payment processing, once you have revenue: roughly 2.9% plus 30 cents a transaction, with no free tier. Before revenue, the biggest cost is usually tool subscriptions you did not need.
Should I self-host to save money?
No, not while you are one person. The savings are small and you pay for them in maintenance, backups and outages at inconvenient hours.
When do I outgrow free tiers?
Hosting usually first, at steady traffic. Database next, as data accumulates. Auth almost never: those free tiers are unusually generous.
How much should infrastructure cost as a share of revenue?
Under 10% is normal and healthy for a small SaaS. If it is much higher, look at what you are storing or processing per user, and at whether your price is too low.
The system behind this, written down
Everything above is the map. The Income Loop is the work inside it: modules 0–6 from the problem you solve to the offer that pays for it, plus ten traffic paths: the deeper post banks, the content sales systems and the full software build sequence, in one place.
Get The Income LoopNot ready to pay for anything? The Basic Income Loop is free and includes a complete seven-day starter for Threads, Instagram or software, enough to find out which one suits you before spending anything.