How Much Does It Cost to Build Custom Software?

There is no price list for custom software, and anyone who gives you a number before asking questions is guessing. But "it depends" is a cop-out, and it is the answer most people get. We build web apps, mobile apps, internal tools, and SaaS products for a living, including Denti, our own multi-tenant dental platform, so we can do better than that: real ranges, the assumptions behind each one, and the levers that actually move the number.

The honest answer, up front

The price depends mostly on what kind of thing you are building. Here are four reference points. The assumptions matter as much as the figures, so read both halves of each.

A simple marketing site or thin MVP. A few pages, a contact form, a CMS so you can edit copy, maybe a waitlist. From an experienced Philippine or Southeast Asian studio, expect roughly $8,000 to $25,000. This is the cheapest real software you can buy, and for a lot of businesses it is all they need this year.

A real web app. User accounts and authentication, a dashboard with actual data, a couple of integrations like payments and email, an admin view, and a database that backs it all. Roughly $30,000 to $80,000. This is the most common thing founders actually mean when they say "an app."

A multi-tenant SaaS. The same as above, but built so many separate customers share one deployment without ever seeing each other's data, plus billing, roles, and the operational tooling to run it as a product. Roughly $80,000 to $200,000.

A cross-platform mobile app. A native iOS and Android app from one codebase, with a backend, accounts, push notifications, and offline handling where it matters. Roughly $60,000 to $180,000 depending on feature depth.

Those are Southeast Asian studio numbers. The same scope from a US or Western European agency typically runs three to four times higher. The spread is rates, not quality: blended senior rates in Manila run $30 to $60 an hour against $150 to $250 in the US, and the engineering is the same Next.js, Postgres, and Flutter either way. We say that plainly because it is the single biggest factor in any quote you will compare, and almost nobody states it out loud.

What actually moves the number

Within any of those bands, six drivers explain most of the gap between the low end and the high end.

Scope and feature count. This is the obvious one and still the biggest. Every screen, every data type, every "and can it also do" is hours. The core of an app is affordable; it is the long tail of features that quietly triples a budget. Ranking features must-have versus later, before anyone writes code, is the single cost decision that saves you the most money.

Integrations. Each external system is real work, not a checkbox. Payments mean a processor, webhooks, and failure handling. Authentication done properly means sessions, password resets, and often social or single sign-on. A third-party API means reading their docs, handling their outages, and dealing with their rate limits. Budget one to three weeks per integration, and be suspicious of anyone who treats them as free.

Web versus mobile, and how the mobile is built. Web is the cheapest surface. Mobile costs more because you are shipping to two app stores and dealing with devices, permissions, and store review. How you build it changes the math a lot. Two separate native apps for iOS and Android roughly double the frontend cost. A single cross-platform codebase in Flutter or React Native lands around 30 to 40 percent over web-only instead of double. If budget is tight and you do not need deep platform-specific features, that one decision saves real money. We dig into the trade-offs in choosing a mobile app tech stack.

AI features. A chatbot demo is a weekend. Production AI is weeks. The gap is everything around the model: feeding it your real data, an action layer that does something with the answer, guardrails for when it is wrong, and a cap on the bill. Denti's AI receptionist books real appointments against live data, remembers the conversation, switches between Filipino and English mid-sentence, detects emergency phrases, and hands off to a human when it should. The model call is the easy 10 percent. If you are weighing AI, what actually works when adding AI to a product is the honest version of that conversation.

Data and compliance. If you handle payments, health data, or personal information at scale, the legal obligations are real and they cost. Encryption for sensitive fields, audit logging, role-based access, and tested backups typically add 10 to 15 percent to a build. Skipping it is the one saving you will regret.

Design polish. A functional interface and a memorable one are different budgets. A standard component library gets you to "clean and usable" cheaply. Custom design, animation, and brand-grade detail are a real line item, which is why our UI/UX work runs as its own 2-to-6-week band. Both are valid choices; just know which one you are paying for.

Phase the build to control spend

You do not have to buy everything on day one, and you usually should not. The pattern that controls spend best is to ship the core, get it in front of real users, and add the expensive features once the product is earning its keep.

That is how we built Denti. Phase one was the core plus the structural bones: scheduling, records, roles, and the multi-tenant skeleton built in from the start because it is cheap at the beginning and a rewrite later. Phase two was automation that drives revenue. Phase three was the expensive conveniences, including the AI receptionist, and it reused the plumbing the earlier phases laid, which is what kept it affordable.

The principle worth copying: sequence by value, and make early phases lay groundwork the later ones reuse. Phasing also de-risks the spend in a way a single big contract cannot, because you get to see working software and decide whether phase two is still worth it before you pay for it.

The ongoing costs people forget

The build price is not the total price. Several recurring costs show up after launch, and a low build with a high monthly bill is a different purchase than it looks.

Hosting. Architecture decides this bill. A well-designed app on a single shared database and a managed platform runs in the low hundreds of dollars a month even with real traffic. A sprawling, over-provisioned setup costs multiples of that for the same load. Ask your vendor what the monthly infrastructure cost will be at launch and at ten times the usage.

Maintenance. Plan on 15 to 20 percent of the build cost per year. That covers dependency updates, security patches, small fixes, and the occasional framework upgrade. Software is not a building that stands once finished; it is a garden. Vendors who quote zero maintenance are deferring it, not eliminating it. For us, maintenance is in the contract on purpose: we stay after launch rather than ship and disappear.

Per-transaction costs. These scale with usage, so they are easy to underestimate at launch. LLM usage if you added AI: usually cents per conversation, but cents without a ceiling become a surprise invoice, which is why we log every model call and alert at 80 and 100 percent of a monthly cap. SMS at volume. Payment processing fees, which are a percentage of revenue forever. Email at scale.

App-store fees. Apple charges $99 a year, Google charges $25 once, and every release costs some engineering time to build, sign, and submit.

How fixed-scope pricing de-risks the whole thing

Most custom software horror stories come from open-ended hourly engagements where the estimate was a vibe and the meter never stopped. The fix is structural, not motivational: a written scope, a fixed price, and a fixed timeline agreed before the work starts.

We run every project this way. Our bands are 6 to 12 weeks for web platforms, 8 to 16 for mobile, 4 to 10 for cloud and conversational AI work, and 3 to 8 for API and integration projects. Fixed scope forces the hard conversations early, in week one: what is in, what is out, what "done" means. The alternative is having them in month four. Changes get priced as new scope rather than quietly stretching the timeline and the bill.

The other half is visibility. We demo working software every two weeks, which we have written about separately, so a project that is drifting gets caught at week two, not week ten. Whoever you hire, insist on both a fixed scope and a recurring demo. Either one alone leaves a gap the budget falls through. You can see the full set of engagement bands and how we scope projects if you want to map your idea to a range.

Red flags in a cheap quote

A $9,000 quote for a "full custom platform" is not a bargain. It is a different product wearing the same name. Things to check before you compare prices:

• A number before any questions. Anyone quoting without asking about your users, your integrations, your platforms, and what "version one" includes is pricing a demo, not your system.
• No line items for staging, migrations, backups, or tests. Production software needs all four. Their absence is exactly how a cheap quote stays cheap, and you pay the difference the first time something breaks in front of a customer.
• "AI included" with no specifics. Ask what the AI actually does with its answer, what happens when it is wrong, and what the monthly model bill is capped at. Vague answers mean a demo bot, not a feature.
• Mobile priced as an afterthought. If iOS and Android together cost less than the web app, ask what is actually being built. Sometimes the honest answer is a website in a wrapper.
• Nothing about handover. No documentation, no admin tooling, no infrastructure-as-code means you are renting, not buying. The cost lands the day you try to switch vendors.

None of these make a vendor dishonest. They make the quote incomparable to one that includes them, and the gap gets invoiced eventually, by someone.

Getting a number you can plan around

Before you commit to building at all, it is worth asking whether you should. For a lot of standard problems an off-the-shelf tool or a no-code setup is faster and cheaper, and we say so when it is true; custom vs. off-the-shelf vs. no-code walks through how to decide. If you do build, picking the right team matters more than the price, and how to choose a software development partner covers what to look for. If you run a clinic specifically, our breakdown of custom clinic software cost has the vertical numbers.

When you are ready for a real estimate, you can cut weeks off the process by showing up with four things: what version one must do versus what can wait, every external system the software has to talk to, which platforms you actually need, and roughly how many users you expect. With that, a competent studio can give you a scoped, fixed price in about a week, and you can finally compare quotes that describe the same product.

That is the conversation we would rather have than any pricing page. If you are seriously pricing a build, send us your scope and we will come back with a real estimate: fixed price, fixed timeline, and an honest recommendation, including "do not build this" when that is the right call.

Frequently asked questions

How much does it cost to build custom software?

+

A simple marketing site or thin MVP runs roughly 8,000 to 25,000 dollars from an experienced Philippine or Southeast Asian studio. A real web app with authentication, a dashboard, and a couple of integrations is roughly 30,000 to 80,000 dollars. A multi-tenant SaaS or a cross-platform mobile app is roughly 80,000 to 200,000 dollars. The same scope from a US agency typically costs three to four times those numbers, because the difference is hourly rates, not engineering quality.

How much does it cost to build a web app versus a mobile app?

+

A web app with auth, a dashboard, and a few integrations is roughly 1,200 to 2,500 engineering hours. Mobile costs more because you ship to two app stores and handle device concerns. A single cross-platform codebase with Flutter or React Native lands around 30 to 40 percent over web-only, not double. Two separate native apps for iOS and Android can genuinely double the frontend cost.

Why are custom software quotes so different from each other?

+

Most of the spread is hourly rate and scope, not skill. Blended senior rates run about 30 to 60 dollars an hour in Manila against 150 to 250 dollars in the US for the same stack. The rest is what got counted: staging, migrations, backups, testing, and handover are real line items that a cheap quote often omits and you pay for later.

What ongoing costs come after the software is built?

+

Hosting, maintenance at roughly 15 to 20 percent of the build per year, and per-transaction costs if you added them: LLM usage for AI features (cap it with alerts), SMS at volume, payment processing fees, and app-store fees. Apple charges 99 dollars a year and Google charges 25 dollars once. Get the monthly running cost in writing before you sign.

How long does it take to build custom software?

+

A small site is a few weeks; a real web app is a couple of months; a SaaS or mobile app is several. We run fixed-scope projects in bands: web 6 to 12 weeks, mobile 8 to 16, cloud 4 to 10, API and integration 3 to 8, and conversational AI 4 to 10. Phasing the build lets you ship the core first and add the expensive features once the product is earning.