TL;DR
Maintaining an app costs 15 to 20 percent of what it cost to build, per year, for maintenance proper: hosting, third-party services, dependency and platform updates, bug fixes, monitoring and renewals. For a $50,000 MVP that is $7,500 to $10,000 a year. The all-in monthly bill that founders actually pay, once you add a small engineering retainer to keep the thing moving, is $2,000 to $5,500 a month for a typical web MVP, $3,000 to $8,000 with a mobile app, and more for an AI product where inference scales with usage.
The number goes wrong in two directions. Founders who budget zero for maintenance discover it in month three as an emergency. Founders who read “plan for two to three times your build cost in year one” are reading a number that includes new features, growth and hiring, which are not maintenance and should be budgeted separately. This post separates them, prices each bucket, and shows what makes the bill double.
What “maintenance” actually includes
Six buckets. Everything a running app costs falls into one of them, and pricing each separately is the only way to get a number you can defend.
| Bucket | What is in it | Typical monthly range for an MVP |
|---|---|---|
| 1. Hosting and infrastructure | Compute, database, storage, bandwidth, CDN, environments | $100 to $600 at MVP scale; $600 to $2,000 once there is real traffic |
| 2. Third-party services | Auth, email/SMS, payments processing fees, file storage, search, maps, analytics, error tracking, uptime monitoring | $150 to $800, mostly usage-based and mostly growing with users |
| 3. Updates and patches | Dependency upgrades, framework major versions, security patches, iOS and Android yearly changes, browser changes, API deprecations from the services in bucket 2 | 4 to 10 engineering hours a month on a web app; double with a mobile app |
| 4. Bug fixes and support | Things that break, edge cases users find, small fixes that are not features | 4 to 12 engineering hours a month in year one, front-loaded |
| 5. Monitoring, backups, incident response | Someone looking at the dashboards, restoring a backup, handling the 2 a.m. outage | Tooling in bucket 2; the people cost sits in the retainer |
| 6. Renewals and fixed fees | Domain, certificates (usually free now), Apple developer program ($99 a year), Google Play (one-off $25), any compliance renewals, an annual pen test if you sell to enterprise | $10 to $50 a month averaged, plus $5,000 to $25,000 a year for a pen test if and when required |
Buckets 1, 2 and 6 are money. Buckets 3, 4 and 5 are engineering hours, and the hours are where the real cost sits: at $75 to $150 an hour for a nearshore or offshore team, or $150 to $250 for US engineers, 10 to 20 hours a month is $750 to $5,000.
What maintenance does not include
Three things get folded into “maintenance” figures and should not be, because they are decisions rather than costs of standing still.
- New features. Anything users did not have at launch. This is the post-MVP roadmap and is budgeted like a build: by scope, not by percentage.
- Growth spend. Marketing, sales, support staff, more engineers. Real, large, and not what “maintain the app” means.
- The rebuild. If the first version needs its core replaced, that is a project with its own number, and when to rebuild your MVP is about deciding it. Maintenance is what you pay so that the rebuild is a choice rather than an emergency.
The often-quoted “two to three times your build cost in year one” is the total of all four. It is a fair warning about the size of the whole undertaking; it is not a maintenance budget.
The number, by MVP type
All-in monthly figures for year one, including a small retainer, at 2026 rates. The retainer assumes 10 to 20 hours a month with the team that built it, at nearshore or offshore rates; US-only rates roughly double the hours line.
| MVP type | Money (hosting, services, fees) | Hours | All-in monthly | Year one |
|---|---|---|---|---|
| Web app, single product, modest traffic | $250 to $1,000 | 10 to 15 | $2,000 to $4,000 | $24,000 to $48,000 |
| Web app plus native mobile app | $300 to $1,200 | 15 to 25 | $3,000 to $6,500 | $36,000 to $78,000 |
| Marketplace or multi-sided product | $400 to $1,500 | 15 to 25 | $3,000 to $7,000 | $36,000 to $84,000 |
| AI product (LLM or model inference in the loop) | $500 to $3,000, scaling with usage | 12 to 20 | $3,000 to $8,000 | $36,000 to $96,000 |
| No-code or vibe-coded MVP | $300 to $1,500 in platform fees | 5 to 15, often by the founder | $800 to $3,500 | $10,000 to $42,000 |
Two readings. First, the low end of each row is a real number for a well-built MVP with a boring stack and a founder who tolerates the occasional rough edge. Second, the “maintenance proper” share of these figures, the 15 to 20 percent of build cost, is the bottom of each range; the top is what happens when the retainer is also quietly doing small features, which it usually is.
A worked year one: a $50,000 web MVP
A B2B web app with an API, built for $50,000 in ten weeks, launched to early customers.
| Line | Monthly | Year | Notes |
|---|---|---|---|
| Hosting (app, database, storage, CDN) | $200 | $2,400 | Managed services, two environments |
| Services (auth, transactional email, error tracking, uptime, analytics) | $250 | $3,000 | Grows with users; payments processing fees are separate and scale with revenue |
| Renewals (domain, Apple program if a companion app follows) | $15 | $180 | |
| Engineering retainer, 12 hours a month with the build team | $1,500 | $18,000 | Updates, bug fixes, on call, small changes |
| Annual dependency and platform pass (framework major, OS releases) | $2,000 | Two days once a year, on top of the retainer | |
| Annual penetration test, if an enterprise deal requires one | $0 to $8,000 | See below | |
| Total, maintenance proper | ~$1,950 | $23,500 to $31,500 | |
| Of which the “keep it alive” minimum (hosting, services, patches, critical fixes only) | ~$900 | ~$11,000 | Roughly 20 percent of the build cost |
That last line is the honest answer to “how much does it cost to maintain an app”: about a fifth of the build per year to keep it alive, about half the build per year to keep it healthy with someone on call and small improvements shipping. A founder who budgets the first number and expects the second experience is the one who feels lied to.
What pushes the bill up
Roughly in order of how often it happens.
A native mobile app. Apple and Google each ship a major OS release a year and deprecate something with it; app store review adds friction to every fix; you now maintain two or three codebases or one cross-platform one with its own churn. Mobile app MVP covers the trade-off; the maintenance consequence is that mobile roughly doubles the hours line.
Integrations. Every third-party API you depend on will change, throttle, re-price or deprecate a version at a time of its choosing. Ten integrations is ten small maintenance projects a year that nobody scheduled.
AI in the loop. Inference is a usage cost that scales with customers, model versions are retired, prompts drift as models change, and the evaluation work to notice that is real engineering time. AI MVP covers the build side; on the running side, budget the inference line separately and re-forecast it monthly.
Technical debt. An app built fast with corners cut costs more to change, so every bug fix and every update takes longer. This is the mechanism by which a cheap build becomes an expensive app; MVP technical debt is about seeing it coming.
No staging environment, no tests, no monitoring. Each of these turns a ten-minute fix into a half-day one, because the engineer cannot see what broke, cannot try the fix safely, and finds out from a customer. Production-ready MVP lists what should be there at launch; each item missing is a maintenance multiplier.
Losing the team that built it. The single biggest jump. A new engineer or agency spends 20 to 60 hours learning the codebase before the first useful fix, and if the handover was poor (no documentation, no access to the accounts, no deployment notes), more. The development contract should specify the handover and a post-launch warranty period precisely to avoid this.
Vibe-coded or no-code foundations. Platform fees replace some engineering hours, but the security and structural fixes described in vibe coding security tend to arrive in year one as maintenance, and no-code products pay a platform rent that rises with usage and cannot be optimised away.
How to keep it down
- Pick a boring stack. Managed database, managed auth, a mainstream framework with a long support window, one cloud provider. MVP tech stack makes the case; the maintenance argument is that boring stacks have fewer surprises per year.
- Ship production-ready. Staging, monitoring, error tracking, backups, a deploy pipeline and dependency scanning at launch cost a few days once and cut the hours line for years.
- Keep the build team on a small retainer for the first six months. Ten hours a month with people who know the code is cheaper than forty hours with people who do not, and it is when most year-one bugs surface.
- Batch the updates. One planned platform pass a quarter, rather than reacting to every deprecation email, halves the interruptions.
- Turn on the free scanners. Dependency and configuration scanning in the code host and cloud account are free and they find the patches before a customer does.
- Do not skip the keep-alive minimum. The founders who spend the most on maintenance are usually the ones who spent nothing for six months and then paid for the emergency.
When the maintenance bill is telling you something
If maintenance proper is above about 30 percent of the build cost a year, and rising, the app is not expensive to maintain; it is badly built, or it has outgrown its first version. Two questions decide which. Is the time going into the same few components every month? That is debt, and MVP technical debt covers paying it down. Is the time going into keeping up with load, new customer types or new integrations the first version never anticipated? That is growth, and it is the scaling or rebuild conversation, which is a good problem to have.
Conclusion
An app costs 15 to 20 percent of its build cost a year to keep alive, and roughly half its build cost a year to keep healthy with someone on call and small improvements shipping. For a $50,000 MVP that is about $11,000 and $25,000 to $30,000 respectively; the “two to three times” figures include features, growth and hiring, and are a company budget, not a maintenance one. Mobile, integrations, AI inference, technical debt and losing the build team are what push it up. A boring stack, a production-ready launch and a small retainer with the people who built it are what keep it down.
If you are budgeting the first year after an MVP, or you have one whose maintenance bill is climbing and want to know whether that is debt or growth, that is a conversation we have regularly as an MVP development company: usually it is a few weeks of paying down the right debt, sometimes it is the next version. Start it here.
Related guides
- How much does it cost to build an MVP
- MVP technical debt
- When to rebuild your MVP
- Post-MVP: what to build next
- How to scale an MVP
- Production-ready MVP
- Scale your MVP
Frequently Asked Questions
How much does it cost to maintain an app?
Maintenance proper, meaning hosting, third-party services, updates and patches, bug fixes, monitoring and renewals, runs 15 to 20 percent of the build cost a year. For a $50,000 MVP that is $7,500 to $10,000 to keep it alive, and $23,000 to $32,000 to keep it healthy with a small engineering retainer. All-in, a typical web MVP costs $2,000 to $4,000 a month in year one; add a mobile app and it is $3,000 to $6,500.
What is the app maintenance cost per month?
$2,000 to $4,000 for a web MVP with modest traffic, $3,000 to $6,500 with a native mobile app, $3,000 to $7,000 for a marketplace, and $3,000 to $8,000 for an AI product where inference scales with usage. The keep-alive minimum, with no retainer and critical fixes only, is roughly $800 to $1,200 a month for a web app.
Is app maintenance really 15 to 20 percent of the build cost?
Yes, for maintenance proper. The figure is a long-standing industry rule of thumb and it holds for MVPs built on a mainstream stack. It does not include new features, growth spend or a rebuild; the “two to three times the build cost in year one” figures include those and are a company budget rather than a maintenance one.
Does a mobile app cost more to maintain than a web app?
Roughly double the engineering hours. Apple and Google ship a major OS release each year with deprecations, store review adds friction to every fix, and you maintain additional codebases or a cross-platform layer with its own churn. Money costs (hosting, services) are similar; the hours line is what grows.
What does an app maintenance retainer include?
Typically a fixed number of engineering hours a month, usually 10 to 20, covering dependency and platform updates, bug fixes, monitoring and on-call response, and small changes. It should not be the vehicle for new features, which are scoped and priced separately. Retainers with the team that built the app are the cheapest hours you will buy in year one.
How much does it cost to maintain an AI app?
The same buckets plus inference, which scales with usage and is usually the largest line: $500 to $3,000 a month at MVP scale and rising with customers. Add engineering time for model version changes, prompt drift and evaluation. Budget inference separately and re-forecast it monthly rather than folding it into a fixed maintenance number.
What are the ongoing costs of a no-code or vibe-coded app?
Platform fees of $300 to $1,500 a month replace some engineering hours, and a founder can often handle small changes. The costs that arrive later are structural: security and permission fixes that need an engineer, platform rent that rises with usage, and a rebuild when the product outgrows the platform. Year one is cheaper; year three often is not.
Can I skip maintenance for the first few months after launch?
You can skip the retainer; you cannot skip hosting, services, security patches and critical fixes, which is the keep-alive minimum of about 20 percent of the build cost a year. Founders who skip everything for six months typically pay more in the emergency that follows than the retainer would have cost.
How do I budget for maintenance when planning an MVP?
Take the build quote, set aside 20 percent of it for year-one keep-alive costs, and another 20 to 30 percent if you want a retainer with the build team for updates and small improvements. Budget new features and growth separately, by scope. Write the post-launch warranty and handover into the development contract so the first months are covered.
When is an app too expensive to maintain?
When maintenance proper is above about 30 percent of the build cost a year and rising. If the hours go into the same components every month, that is technical debt to pay down. If they go into load, new customer types or integrations the first version never anticipated, that is growth, and the answer is the next version rather than more maintenance.
