What does custom software cost in Denmark in 2026?

The short version owners need: Danish studios still often charge 900–1,500 DKK an hour. That line hardly moved. What moved is how much comes out of the hour. Boilerplate, screens, tests, API stubs and the tenth pass on a form — the work that used to fill the timesheet — got radically cheaper to do. That is why a bounded custom system can sit at 18,000–54,000 DKK in 2026 and be the project that used to come back at 150,000.

What “6× more for the same price” actually means

Six times is not a brochure number. It is the working model we price from in 2026: the part of the project that is *building* — screens, flows, tests, repetition, the boring glue between systems we already know — takes a fraction of the time it took in 2022. If you used to buy four well-built features for 36,000 DKK, you can now buy something like 24. That is the medium package. Not a different product. Same price. More system.

That is why this article is no longer just a list of “what does an app cost”. The price question moved. The 2026 question is: *how much system can you get before the budget runs out — now that the code is no longer the expensive part?* English package prices on our pricing page are in EUR. This guide uses DKK because that is how the Danish market still quotes.

What does custom software cost when the developer is 6× faster?

The old ranges still come from public 2025–2026 agency guides: websites from about 5,000 DKK, heavy custom sites, apps 50,000–500,000 DKK, SaaS MVPs 150,000–350,000 DKK. What changed in 2026 is the middle. A bounded SME system no longer has to start at 80,000 DKK just because someone multiplied 80 hours by 1,200. Two “web apps” can still cost five times each other if one has MitID and e-conomic and the other is three screens and a form. AI does not remove that gap. It makes it obvious.

What “custom” means — and what it does not

Custom software is a system built for your process. Not a WordPress theme with a plugin. Not an Excel file with macros. Not ChatGPT with a login in front. And not a shelf product you bend with 14 workarounds.

It is the right choice when you lose hours every week typing the same data into two systems, when your process does not fit a standard CRM, or when you need Danish keys: MitID, MobilePay, e-conomic, Dinero, PostNord. And when you want to own the code instead of paying per seat for life.

It is the wrong choice if you actually need a website with text and a contact form. Then buy a website. See websites. AI does not turn a brochure into a system — and it should not.

What 18,000, 36,000 and 54,000 DKK buy now

Some houses show packages. That is still rare. It is also what makes comparison possible when the hourly rate no longer tells the truth. Ceptiv’s own SME packages, August 2026:

Read that again: 12, 24 and 36 features. That is not “an app shell and a promise”. That is the volume the AI shift made payable in the SME band. A written quote within 24 hours. See pricing.

Other Danish houses still sell other products: sprints from about 35,000 DKK per two weeks, discovery first around 25,000 DKK, classic T&M at 200 hours × 1,200 DKK = 240,000 DKK. Ask what you get *out* of those hours in 2026. If the answer is the same as in 2023, you are paying for a world that is gone.

What AI does not make six times faster

If everything became 6× cheaper, custom software would cost 3,000 DKK. It does not. Because a chunk of the price never sat in the keyboard.

1. Your decisions

“We’ll know it when we see it” is still not a project. It is time and materials. AI can show you three versions of a screen in an afternoon. It cannot choose whether you are B2B or B2C, whether invoices go to e-conomic or Dinero, or whether the customer may see the price before login. Every week you hesitate costs calendar — not because the code is slow, but because you are.

2. Integrations nobody has shipped before

A module the vendor has shipped before is exactly where 6× lands. The first time anyone talks to a moody API, it is still real work. That is why agencies talk about “40 ready integrations”. It is the only reason an SME package can sit at 18,000 DKK instead of 80,000 DKK.

The deep guide: e-conomic and Dinero in your own system. MitID is its own story: MitID in your own app.

3. Messy data and real people

Old Excel is never ready to import. An auditor who may see postings without CPR is a different project from “everyone can do everything”. Permissions are invisible in a pitch deck and expensive in QA. AI does not write your test data. It also does not clean 8,000 crooked customer names.

4. Operations

Build is the one-off. Operations is what you pay so the system does not rot: hosting, database, SSL, backups, monitoring, security updates, bug fixes and a little ongoing support. A system with no ops budget is a system you administer yourself. Most people notice on the day the certificate expires. AI did not make servers free.

5. Design that has to be yours

A clean interface on a design system is included in most sane packages — and it got faster to keep consistent. A unique brand universe with animation and illustration is still an add-on. Most SMEs should pick the first and spend the 6× gain on more flows, not more gradients.

Fixed price, hourly, or sprint — after the shift

• Fixed price fits when you can describe the result: “a customer login, three screens, invoices go to e-conomic, MobilePay at checkout.” It fits more often now, because you can buy *more* result without opening the timesheet.
• Hourly fits when you are discovering the product: new SaaS, uncertain behaviour, you change direction every week. Be careful: hourly on 2022 estimates in 2026 is the most expensive way to buy the AI gain — you pay the old volume of hours at the new speed.
• Sprints fit when you want to control pace and can stop after each delivery — but you must hold the direction yourselves. Ask how much actually comes out of two weeks now.

Fixed price is not magic. Scope is fixed when the price is fixed. New wishes are an add-on. That is fair if it is in the agreement from day one. The difference is that the add-on is often a week now — not a quarter.

Hidden costs most people still forget

1. Your own time. Workshops, test data, decisions. Budget 4–8 hours a week during the build. AI did not make *you* 6× faster at answering email.
2. Data cleanup. Old Excel is never ready to import.
3. Licences you already pay (e-conomic Pro, Dinero Pro, HubSpot seats).
4. MitID transactions if you use a broker. The core is øre per login; the broker adds its own layer. MitID in your own app.
5. A rewrite in 3 years if you buy a system only one freelancer understands — or a system only one AI chat understands, and nobody reviewed.

When 18,000 DKK is realistic — and when it is not

Realistic if you accept one core flow, one accounting system, a design system (not an art project), decisions within 24–48 hours, and you will not mid-project “also want an app”. 18,000 DKK became *more* realistic, not less, because the 12 features can actually be reached.

Not realistic if you say “it should do what Neutral or Billy does, just ours”, “we don’t know if this is B2B or B2C”, “can you also take over our old Access database?”, or “we need AI that understands the whole business”. AI in an SME system is usually one concrete feature — a draft, a classification, an assistant on one screen — not a staff member you can fire. AI integrations.

How to read a quote in 2026

Ask for these five lines. If they are missing, the quote is not finished. If it looks like a 2023 quote with 180 hours and three screens, it is finished — you just should not buy it.

1. What is not included
2. Which integrations, and whether they have been built before
3. What the monthly fee covers — and what costs extra
4. Who owns the code and the data
5. What happens if you stop after 12 months

A 24-hour quote is only valuable if it is specific. A vague “80–200,000, we’ll see” is a sales meeting, not a quote. And a quote that still sells you *hours* as if AI did not happen is a quote for the old efficiency at the new rate.

What to do now

Write the three flows you hate most in a normal week. Write which systems you already pay for (e-conomic, Dinero, HubSpot, MobilePay). Do not write “can you just use AI”. Write what has to work on Monday. That is enough of a brief. If you want a fixed number and a date, send it — and expect more system than the same money would have bought three years ago.