BareStack
← Blog
guides

How Much Does a Custom Web App Really Cost in 2026?

Custom web app cost in 2026, with real price bands, published agency hourly rates and UK day rates, plus what pushes a quote up before you sign.

·11 min read
How Much Does a Custom Web App Really Cost in 2026? — illustration

TL:DR: A custom web app in 2026 costs about £8,000 to £40,000 for a single-workflow internal tool, £35,000 to £140,000 for a client-facing app with accounts and payments, and £90,000 or more for a multi-tenant product. The total is a rate multiplied by a day count, and both halves are published: Clutch puts most agency work at $25 to $49 an hour, while UK public sector rate cards list a developer at £1,405 a day. That three to four times gap for identical work is overhead, project management and risk, not better code. What you are really buying is who carries the scope risk and who is still answering emails in month eight.

Two agencies quote the same brief. One says £14,000, the other £68,000. Neither is lying, and neither number is the price of code. One is selling twenty days of one developer. The other is selling forty days, a project manager, a designer, a QA pass, and an insurance policy against the thing not working. Compare them by asking for the day count behind each total.

What "custom web app cost" actually means in 2026

A custom web app has no list price. It has a build cost, a running cost and a maintenance cost, and most buyers only ask about the first. The build is a rate multiplied by a day count, the running cost is a server and a database, and the maintenance cost is the hours somebody spends on it after launch. That last line ruins budgets because nobody puts it in the first quote.

Clutch's web development pricing guide, updated on 21 September 2026, puts most agencies at $25 to $49 an hour, with US, Canadian and Australian firms at $100 to $149, India below $25, and the United Kingdom listed as unknown. Its average reviewed project is $66,499 over roughly nine months, per Clutch's pricing guide. That average is pulled up by a few large builds, which is why the same page also says most projects come in under $10,000.

The UK picture is stranger. Private agencies do not publish rate cards, so the only transparent UK rates are public sector ones. Deloitte's G-Cloud 15 standard rate card, published in January 2026, lists £1,275 a day at the HEO/SEO grade, £1,530 at Grade 7, and a named software developer role at £1,405, per Deloitte's G-Cloud 15 rate card. Those are ceiling prices from a supplier carrying overhead, insurance and a bid team, so treat them as the top of the UK market rather than the middle.

BuildEstimated daysAt $49 an hour ($392 a day)At £1,275 a day
Single-workflow internal tool20 to 35$7,840 to $13,720£25,500 to £44,625
Client app with accounts and payments60 to 110$23,520 to $43,120£76,500 to £140,250
Multi-tenant product, version one140 to 250$54,880 to $98,000£178,500 to £318,750

Day counts are v1 scope estimates with no legacy integration. The rates are published numbers read in November 2026, assuming eight-hour days. A 60-day build is $23,520 at the top of Clutch's all-location band or £76,500 on a UK public sector card, for the same work. If a quote sits far outside both ends, one of you has misunderstood the day count.

Option 1: A freelancer at $25 to $49 an hour, the cheapest way in

A solo developer is the cheapest credible way to get a working application. At $49 an hour a 25-day internal tool is under $10,000, and Clutch's India band sits below $25, so the same screens can cost a fraction of a large supplier's day rate.

What you give up is cover. One person is one holiday, one illness, one contract that ends mid-sprint. You also inherit their documentation habits, which are often none, and a codebase only they understand. That is survivable for a tool with three users. It is expensive for anything touching payments, because the next developer spends a week learning what the first one did.

Ask for three things before signing: the repository in your own GitHub account from day one, a staging environment that is not their laptop, and a written deploy process you could follow without them. If any of those is refused, the hourly rate is not the number that matters.

Pick a freelancer if the scope is one workflow, the users are internal, and you can write a spec in two pages. Skip it if the app will hold client data, take payments, or outlive the person building it.

Option 2: A small agency on a fixed scope, the quote that survives month three

A small agency is the middle of the market and the hardest to compare, because the day rate hides the team shape. Clutch's $66,499 average sounds large until you notice that a three-person team at a modest blended rate gets there in eight weeks. That figure benchmarks the projects agencies choose to list, not your app.

The useful question is what is included at that rate. A fixed-scope quote with a written change-request process is worth more than a cheaper time-and-materials quote with neither, because the second transfers all scope risk to you. Scope creep is the default outcome of a brief that says "and a dashboard" without saying which numbers are on it.

Ask for the day count split by role, the change-request rate, and what happens to the code if you stop paying. A serious agency answers all three in writing on the first call, because they have been asked before.

Pick an agency if the app is client-facing, needs design work, and somebody must be accountable when a deadline slips. Skip it if you cannot describe the workflow in a paragraph, since a fixed price on an unclear brief is a deposit on an argument.

Custom web app cost breakdown with budget sheets and a pen

Option 3: A large supplier at public sector day rates, the expensive way to buy certainty

The top of the market is a supplier with a published rate card, and the UK public sector is where you can read one. Deloitte's G-Cloud 15 card prices a software developer at £1,405 a day, a senior developer at £1,630 and a lead developer at £2,020, per the G-Cloud 15 rate card. Forty days of one mid-level developer is £56,200 before a project manager, a designer or a tester is added. That is the price of a supplier that can pass a security review, carry professional indemnity insurance and staff a bid team.

This tier exists for buyers whose real problem is procurement risk rather than code. Regulated industries and public bodies are paying for the paper trail as much as the software, and a five-person business is subsidising overhead it will never use. There is a second cost too: pace. Change control turns a two-line copy change into a ticket, and the invoice grows with the calendar rather than the work.

Pick a large supplier if you need security accreditation, formal SLAs, or a contract your legal team can defend. Skip it if the app is internal and the worst outcome is a spreadsheet for another month.

Option 4: A no-code platform, cheapest to start and hardest to leave

A no-code build changes the shape of the cost rather than removing it. You pay a small subscription and swap developer days for your own time or a consultant's. Bubble runs $32 a month on Starter, $119 on Growth and $349 on Team when billed annually, with workload allowances of 175,000, 250,000 and 500,000 units a month, per Bubble's pricing page. Retool prices internal tools at $10 per builder and $5 per internal user a month on Team, billed annually, per Retool's pricing page. A 25-person ops team pays $1,620 a year there, about one day of UK public sector developer time.

The trap is not the subscription, it is the ceiling. Workload units and per-seat pricing both scale with success, so the app that costs $32 a month at launch costs $349 when the team grows, and the platform decides which features sit in which tier. Leaving is the expensive part: your logic lives in somebody else's editor, and export gets you data, not the application.

No-code is right more often than agency owners admit. Under ten users, with requirements that will change next quarter, a $349 a month platform fee and three weekends of work beat a £30,000 build, and get there a year sooner.

Pick no-code if the tool is internal, the user count is small, and you could rebuild it elsewhere in a fortnight. Skip it if the app is the product, or if per-seat pricing at your expected headcount costs more than a build would.

How to choose, the decision tree

Four variables decide this: who uses the app, how many of them there are, how much the requirements will move, and what happens if it breaks.

  • If the app is one internal workflow under ten people, start on no-code at $32 to $349 a month and revisit in a year.
  • If the users are external and the app takes money, you need accounts, permissions and a payment flow a platform will not cover cleanly. Budget 60 to 110 days.
  • If the requirements are stable and you have a two-page spec, a freelancer at $25 to $49 an hour is the best value in the market, provided the repository is yours.
  • If the app is client-facing and somebody must answer when it breaks, a small agency on fixed scope is worth the premium over a freelancer.
  • If your buyer requires accreditation or an SLA, use a supplier with a published rate card and accept £1,405 a day as the floor.
  • If the app is the product rather than a support tool, skip every shortcut and budget 140 to 250 days. Multi-tenant, billing and permissions are where cheap builds die.
  • If you cannot say who maintains it in year two, stop. That question costs nothing now and everything later.

The realistic answer for most small businesses is a fixed-scope build of 40 to 60 days at a blended rate, with hosting and maintenance separated in the contract. That lands between £18,000 and £45,000 depending on who does it, and it is the band where the work is defined well enough to price.

Integrating with the rest of your stack

Hosting should stay cheap. A Hetzner cloud server starts at €5.99 a month excluding VAT, per Hetzner's cloud pricing, and a managed Postgres and auth layer starts at $25 a month on Supabase Pro, per Supabase's pricing page. If a quote includes £400 a month of cloud infrastructure for 30 users, ask what is running on it.

The expensive part is the seams. An app that cannot see your CRM duplicates every customer record, and one that cannot invoice turns a two-minute job into an evening. Wire the app to the tools you already run rather than rebuilding them, and if you do not have a CRM, BareStackOS is free and open source with no seat limit, which removes one subscription before the build starts.

Ask the exit question before you sign: if you fell out with the developer tomorrow, could you take the repository, the database and the deploy process and hand them to somebody else on Monday? If the answer needs a meeting, the price is wrong. Our comparison of build approaches sets out how fixed-scope work differs from subscription tools, and the BareStack FAQ covers what happens after handover. For hosting, this Hetzner cost comparison is the honest starting point, and the $5 a month website guide shows how low running costs go. For the same arithmetic from the buying side, read open source versus SaaS and the true cost of a free SaaS plan.

Bottom line

A custom web app in 2026 costs between £8,000 and £40,000 for an internal tool, £35,000 to £140,000 for a client-facing app, and £90,000 upward for a product. The published rates behind those bands are $25 to $49 an hour for most agencies and £1,275 to £1,630 a day on a UK public sector card, and the difference is overhead and risk transfer rather than code. Price the day count, not the total.

FAQ

How much does a custom web app cost in 2026?

Most custom web apps land between £8,000 and £140,000 depending on scope. A single-workflow internal tool runs 20 to 35 developer days, a client-facing app with accounts and payments 60 to 110, and a multi-tenant product around 140. At the published extremes of $392 a day and £1,275 a day, the same 60-day build is $23,520 or £76,500.

Why do two quotes for the same app differ by five times?

Because they price different day counts and different risk. A £14,000 quote is usually one developer for twenty days with no design, QA or project management. A £68,000 quote is a team for forty days with all three, plus a warranty. Ask for the day count split by role and the gap usually becomes explainable.

Is a custom web app cheaper than no-code?

Over one year, no. Bubble costs $32 to $349 a month and Retool charges $10 per builder and $5 per internal user, so a small internal tool runs under $2,000 a year on either. Over three to five years at 50 or more users, per-seat pricing can pass a one-off build.

How long does a custom web app take to build?

A 20 to 35 day internal tool takes four to seven weeks once work starts. A 60 to 110 day client app takes three to five months, and a multi-tenant product six to nine. Clutch's reviewed projects average about nine months end to end.

What does it cost to run a custom web app each month?

Hosting starts at €5.99 a month on a Hetzner cloud server, with dedicated plans from €43.49, or $25 and up for a managed database and auth service. Maintenance is the real line: one to three developer days a month for updates, backups and fixes, which at £1,275 a day is £1,275 to £3,825 a month if you buy it in. Get that number in writing before launch.

Should I hire a freelancer or an agency for a web app?

Hire a freelancer for one workflow, internal users and a spec you can write in two pages, at $25 to $49 an hour. Hire an agency when the app is client-facing, needs design and QA, and somebody must be accountable for a deadline. The deciding factor is who answers when the only person who understands the code is away.


Photo by Austin Distel on Unsplash

Photo by NORTHFOLK on Unsplash

About the author

Anirudh Prashant · Founder & Lead Engineer, BareStack

Founder of BareStack. Builds custom, no-bloat software, self-hosted tooling, and AI automations for solopreneurs and small teams.