What Your SaaS Should Cost to Host at Every Revenue Stage

TL;DR — Your infrastructure should cost 5-15% of revenue. At $0 MRR, spend $0. At $1K MRR, spend $50-150. At $10K MRR, spend $500-1,000. We built a fictional SaaS called TaskFlow and tracked its hosting costs from launch to 15,000 users — infrastructure never exceeded 3% of revenue. If yours is above 20%, you're overspending. If you have no backups and no staging env, you're underspending.

The golden rule: infrastructure should be 5-15% of revenue

Most founders either overspend on infrastructure they don't need yet or underspend until something breaks. There's a simple benchmark that works at every stage.

Infrastructure cost as a percentage of MRR should land between 5% and 15%. Below 5%, you're probably cutting corners that will cost you later. Above 15%, you're burning money on resources you haven't grown into.

This isn't a guess. It's the range we see across hundreds of SaaS apps running on RaidFrame, and it aligns with industry benchmarks from Bessemer, a16z, and Scale VP.

How much should you spend at $0 MRR?

$0-20/mo. Ideally $0.

You have no revenue. You have no users. You have no business spending $50/mo on infrastructure for an app that doesn't exist yet.

You might be coming from Render, Railway, Neon, or Supabase — all have free tiers. But RaidFrame's free tier includes app hosting, Postgres, and SSL with no cold starts and no pausing after inactivity. Ship your MVP, get feedback, worry about infrastructure later.

What does $1K MRR infrastructure look like?

Budget: $50-150/mo (5-15% of revenue)

You have paying users. They expect the app to work. This is where your stack gets real.

On RaidFrame: $21-28/mo covers all of this. App container ($7-14), managed Postgres ($7), Redis included, background jobs included. See pricing.

The mistake at this stage is adding monitoring tools that cost more than your hosting. You don't need Datadog at $1K MRR. Built-in logs and a free Sentry tier are enough.

What changes at $5K MRR?

Budget: $250-500/mo (5-10% of revenue)

You're past early traction. Users depend on you daily. Downtime has a real cost — churn, refunds, support tickets.

On RaidFrame: $60-150/mo. Auto-scaling included on Pro plans. Staging environments at standard per-service pricing. Your bill scales with what you provision, not with traffic.

When should you add a staging environment?

When your first customer emails you about a bug you shipped. For most teams, that's somewhere between $3K and $5K MRR. The staging env costs $20-50/mo. The alternative is breaking production and losing customers.

What does $10K MRR infrastructure look like?

Budget: $500-1,000/mo (5-10% of revenue)

At this stage, you're thinking about multi-region deployment, dedicated resources, and compliance.

On RaidFrame: $150-400/mo. Multi-region deployment and managed backups included on Pro. Compliance tooling available on Enterprise.

What about $50K+ MRR?

Budget: $2,500-5,000/mo (5-10% of revenue)

Full production stack. SLA guarantees. Dedicated support. You're not optimizing for cost anymore — you're optimizing for reliability and speed.

At this stage, your infrastructure includes dedicated database clusters, multi-region with automatic failover, full observability pipelines, disaster recovery with tested runbooks, and a support channel with a human who knows your stack.

If you're spending more than $5,000/mo at $50K MRR, something is wrong. Either you're over-provisioned, running zombie resources, or paying for enterprise tiers you don't need. Cut your bill by 40% without sacrificing performance.

Case study: TaskFlow — from $0 to $50K MRR

TaskFlow is a fictional project management SaaS. One founder, Node.js API, React frontend, Postgres database. Here's exactly what they spent on infrastructure at each stage.

Months 1-3: Building ($0/mo)

Pre-revenue. Free tier on RaidFrame. One container, managed Postgres, SSL. Zero dollars.

MRR: $0 | Infra cost: $0/mo | Infra as % of revenue: N/A (free tier)

Month 6: First traction ($29/mo)

200 users, 50 paying at $19/mo. Upgraded to always-on container, added Redis for session management.

MRR: $950 | Infra cost: $29/mo | Infra as % of revenue: 3%

Month 12: Growing ($79/mo)

2,000 users, 400 paying at $19/mo. Added auto-scaling, staging environment, automated backups. Moved to Pro plan.

MRR: $7,600 | Infra cost: $79/mo | Infra as % of revenue: 1%

Month 24: Scaling ($450/mo)

15,000 users, 3,000 paying at $19/mo. Multi-region deployment, read replicas, CDN, full monitoring stack. Hired a second engineer.

MRR: $57,000 | Infra cost: $450/mo | Infra as % of revenue: <1%

The pattern

Infrastructure as a percentage of revenue stayed under 3% at every stage and dropped as revenue grew. TaskFlow never had a surprise bill. They never had to re-platform. They scaled up their RaidFrame resources as revenue justified it — and infrastructure never came close to eating their margins.

That's what predictable infrastructure looks like. Start on the free tier and scale without re-platforming.

How do you know you're overspending?

These are warning signs that your infrastructure budget is out of control:

• Infrastructure exceeds 20% of MRR. At $5K MRR, your hosting should not be $1,500/mo. Period.
• You're paying for resources you don't use. That 8GB instance running at 15% CPU utilization. The staging database nobody's touched in 3 months.
• You have per-seat fees stacking up. Platform fees, monitoring seats, error tracking seats — $25/seat/mo across 4 tools for a 3-person team is $300/mo in seat tax alone.
• You're running on AWS "because everyone does." Self-managed AWS requires 10-20 hours/mo of ops time. At $75/hr, that's $750-1,500/mo in hidden labor. That's the real cost.

How do you know you're underspending?

Equally dangerous, and more common than founders admit:

• No automated backups. If your database dies and you lose everything, you didn't save money — you lost your company.
• No staging environment. You're testing in production. It's a matter of when, not if, you ship a breaking change to paying customers.
• Frequent downtime. If your app goes down weekly because your single container runs out of memory, spending $7 more per month on a bigger instance is not optional.
• No monitoring or alerting. Your users shouldn't be your monitoring system. If you find out about outages from support tickets, you're underspending.

FAQ

What percentage of revenue should go to infrastructure?

5-15% is the healthy range. Early-stage SaaS typically runs 8-12%. If you're above 20%, audit your resources. If you're below 3%, make sure you have backups and monitoring.

How much should a pre-revenue SaaS spend on hosting?

$0-20/mo. Use free tiers. Do not spend money on infrastructure before you have paying users. RaidFrame's free tier includes everything you need for an MVP.

When should I upgrade from free tier hosting?

When you have paying users who expect reliability. Free tiers have cold starts, pausing behavior, and resource limits that are fine for development but unacceptable for production. That's typically around 50-100 paying customers.

Is it worth paying for a staging environment?

Yes, once you have paying customers. A staging env costs $20-50/mo. One production bug that causes churn costs far more. It's the cheapest insurance in your stack.

How do I know if I'm over-provisioned?

Check actual CPU and memory utilization. If your average CPU is under 30% and memory is under 50%, you're over-provisioned by at least 2x. Most apps are. Right-size quarterly.

Should I use AWS or a PaaS at $10K MRR?

At $10K MRR, your time is worth more than the marginal savings of self-managing AWS. A PaaS like RaidFrame gives you the same capabilities — multi-region, auto-scaling, managed databases — without the 10-20 hours/mo of ops overhead.

How does RaidFrame keep infrastructure costs predictable?

Flat per-service pricing. No per-seat fees. No bandwidth surcharges. No surprise bills from traffic spikes. You pay for the resources you provision, and your bill scales linearly with your decisions.

Related reading

• How Much Does It Actually Cost to Host a SaaS?
• Cloud Cost Optimization: Cut Your Bill by 40%
• Your Hosting Bill Should Scale Linearly
• Multi-Region Deployment Strategy
• Deploy Next.js + Postgres for Free