Every founder asks the question at some point: should we self-host this open-source tool, or pay for the SaaS?
The honest answer is "it depends" — but most founders use the wrong framework to decide. Here's the real 5-year cost picture.
The fake math (what most blogs tell you)
"Self-hosted is free, SaaS costs $X/month, therefore self-hosted wins."
This is wrong. Self-hosted isn't free. It costs:
- VPS ($5-100/mo per tool)
- Setup time (4-40 hours initially)
- Maintenance (~2 hours/month per tool)
- Downtime (~99.5% vs SaaS's 99.95% = 4 hours/year of outages you have to fix)
- Security patches (~1 hour/month if you're disciplined)
- Backup management (when it fails, you pray)
- Scaling pain (~5 hours when your VPS runs out of RAM)
The real question: is the dev time worth more than the SaaS subscription?
The real 5-year math
Let me show you a real-world calculation. Same team, same tools, two paths.
Scenario: 10-person SaaS startup
SaaS path (year 1):
| Tool | Cost/month |
|---|---|
| Notion (team) | $100 |
| Slack | $125 |
| Calendly | $150 |
| Mailchimp | $300 |
| Linear | $80 |
| HubSpot CRM | $450 |
| Zoom | $150 |
| Google Workspace | $180 |
| TOTAL | $1,535/mo |
Year 1 SaaS cost: ~$18,420
5-year SaaS cost: ~$92,100 (assuming no price hikes — which is unrealistic; expect 30-50% growth)
Self-hosted path
Year 1 setup:
- 1 weekend (~20 hours) for one engineer to deploy everything
- @ $50/hour founder time = $1,000 (opportunity cost)
Year 1 ongoing:
- 3 VPS ($30/mo total) = $360/year
- 4 hours/month maintenance × 12 = 48 hours × $50 = $2,400/year
- Email sending (AWS SES at 1M emails) = $100/year
- Year 1 total: ~$3,860
Year 2-5 (no setup cost): $2,860/year × 4 = $11,440
5-year self-hosted cost: ~$15,300
The verdict
Save ~$76,800 over 5 years by going self-hosted.
But wait — there's nuance.
When SaaS actually wins (the honest version)
There are real scenarios where SaaS is the right call:
1. Pre-revenue / pre-PMF stage
If you're solo or 2-3 people building toward product-market fit, your time is your only asset. Spending 20 hours self-hosting tools that don't help you ship product is bad math.
Rule: if your runway is < 12 months or you don't have product-market fit, don't self-host. Pay the SaaS tax. Optimize cost later.
2. Non-technical teams
If your team isn't technical, self-hosting is impossible. The "free" tools become very expensive when nobody can fix them.
3. Compliance requirements
Sometimes SaaS is required by compliance (SOC 2, HIPAA, FedRAMP). Open-source tools without certifications can't be used in regulated industries.
4. Tools used by 100s of customers
If you're going to give external users access (e.g. a customer portal), SaaS reliability matters more. Internal tools have lower stakes.
When self-hosting wins clearly
1. You have 10+ employees
The math flips hard at 10+ users. SaaS pricing scales linearly with users; self-hosted is flat. At 50 employees, the math is overwhelming.
2. You're privacy-sensitive
GDPR, DPDP (India), HIPAA — once your industry forces data sovereignty, self-hosting becomes mandatory, not optional.
3. You're profitable
Unprofitable startups should burn cash on SaaS to move faster. Profitable startups should reinvest in efficiency. Self-hosting is reinvesting your own money to lower your cost basis.
4. The tool is a margin killer
If a SaaS tool costs more than 5% of your gross margin, self-host it. For example, Twilio at $0.0075/SMS at scale eats margin. Self-hosted alternatives exist.
The hybrid strategy (what most smart teams do)
Don't go binary. The right answer is usually:
| Category | Pick |
|---|---|
| Mission-critical for users | SaaS (e.g. Stripe, AWS) |
| Internal team productivity | Self-host (Notion → AppFlowy, etc.) |
| Marketing tools | Self-host if technical, SaaS if not |
| Auth + payments | Always SaaS (don't roll your own) |
| Email sending | SaaS infrastructure (AWS SES, Mailgun) + self-hosted UI (Listmonk) |
| Analytics | Privacy-first OSS (Plausible) — costs less anyway |
The 80/20 rule: identify the 3-5 most expensive SaaS tools you use. Self-host those. Pay for the rest.
Real cost factors most people miss
Hidden cost of SaaS lock-in
- Migration cost: when prices triple in year 3, switching costs you 2-4 weeks
- Data portability: most SaaS make export deliberately painful
- Feature deprecation: vendor removes a feature you depend on
- Acquisition: vendor gets bought, product becomes worse
- Outage: their downtime is your downtime, with no recourse
These don't show up in the monthly bill. They show up at exactly the wrong moment.
Hidden cost of self-hosting
- Skill required: you need at least one team member who can debug Linux/Docker
- Backup discipline: if you forget backups, you lose data when the VPS dies
- Security patches: an unpatched OSS instance is a security risk
- Time of recovery: when something breaks, no support ticket — you fix it
These don't show up in the monthly bill either. They show up the night before a launch when your VPS runs out of disk space.
My recommendation
For founders pre-PMF: Pay for SaaS. Ship product. Don't optimize cost.
For founders post-PMF with technical team: Audit your subscription stack. Replace the top 3-5 most expensive with self-hosted equivalents. The savings fund a new hire.
For founders running side projects: Self-host everything. The setup time becomes learning time. The cost is your hourly rate × time saved.
The full curated picks
If you're ready to self-host, **StackPicks** has the **160+ open-source tools** and **13 ready-to-ship stack bundles** curated by builders for builders.
Or browse **SaaS alternatives** — every page has a top-3 OSS replacement for a specific paid tool.