What custom software actually costs in 2026

A quote arrives for a system you need and the number has six digits in it. The instinct is to assume software is expensive. The more useful question is what, specifically, you are being charged for.

The line items nobody itemises

Take a typical agency or development shop engagement and separate the parts:

• Discovery. Three to four weeks of workshops and stakeholder interviews producing a requirements document. Some of this is genuinely necessary. Much of it is reconstructing knowledge that already exists in your head and in your spreadsheet.
• Design. Wireframes, then mockups, then revisions to the mockups, then sign-off. Four to six weeks, most of it spent making a picture of software so that a different person can build it later.
• The handoff. Not a line item, but the reason for several. A designer hands to a developer. What was ambiguous in the mockup gets resolved by guessing, and the guesses come back as change requests.
• Project management. A person whose function is to move information between people who are not talking directly.
• Account management. A person whose function is to move information between you and the project manager.
• The build. Where the software gets written. Often by whoever is cheapest on the bench, because the senior people are on the pitch for the next client.
• Padding. Every handoff above is a place the project can go wrong, and the estimate has to cover the possibility that it does.

Add it up honestly and the writing of the software is a minority of the invoice.

What changed, precisely

Two things, and it is worth being exact about which.

The first is that writing code got dramatically faster. That is real, and it is the part everyone talks about. It compresses the build.

The second matters more and gets almost no attention. When one person can hold the whole thing, the handoffs stop existing. There is no designer-to-developer translation loss, because it is the same person. There is no project manager moving information between them, because there is nobody to move it between. There is no padding for coordination risk, because there is no coordination.

Removing the handoffs removes the phases that existed to manage the handoffs. That is where the number actually falls.

What has not changed

The cost of deciding what to build did not move at all.

Understanding how your operation actually runs, which parts are load-bearing, what the edge cases are, and what should deliberately be left out is the same work it always was. It takes a conversation with someone who asks good questions, and it cannot be automated, because the information exists only in the heads of people who have never had to write it down.

Anyone selling you the second thing at the price of the first is selling you something else. Usually a template.

The arithmetic, roughly

A conventionally quoted custom project runs about thirty-three weeks: four weeks discovery, six design, sixteen build, four QA, three handover. The equivalent work, done without the handoffs, runs about five: two days of conversation, three days producing a written scope, three weeks building, one week of you using it against real work while things get fixed.

The compression is not people working harder or faster. It is phases that stop being necessary when the person scoping the work is also the person building it.

What to ask for instead of a quote

A number without a scope attached is meaningless, and a scope written after you have paid a deposit is a negotiating position. What you want, before any money moves:

• What gets built, named specifically enough that you could tell whether it was delivered
• What is deliberately out of scope, which is the more informative half
• What it costs, fixed, and what would change it
• Who owns the code, the accounts, and the data on the day it ships
• What happens if you want to leave

That last one is the tell. If leaving is expensive, the pricing is not really about the build. It is about the years afterwards.