Custom software budget and timeline estimator

How the bands are decided

The tool works in two steps. First the points of the modules you ticked are added up; then the total is mapped onto one of the three published bands. The mapping rule is fixed and fully visible here:

Sites are added on top: a single site adds nothing, 2-3 sites add 1 point and 4 or more add 2. Running the same process in several places means inter-site transfers, per-site permissions and consolidated reporting, so it is genuine scope rather than a bigger licence.

The points follow one logic: a module’s weight is set not by its screen count but by the data model and the number of rules behind it. An approval flow adds a status field to a single table, so it scores 1. Production planning with bills of materials brings five or six interlinked tables, versioning and derived fields, so it scores 3. An AI assistant needs a separate index, embedding generation, source attribution and accuracy testing, so it also scores 3. This ranking comes from our own delivery experience; we do not claim it is a measured industry standard.

Why the user count does not change the price

The form asks for a user count but it has no effect on the band — and that is the model, not an oversight. At KodDelta the licence is perpetual and the user count is unlimited: five people and five hundred people cost exactly the same. We still ask, because infrastructure sizing and the training plan need the number during discovery. With rented software the opposite is true; there the cost is multiplied directly by headcount. To see that difference in figures, use the 5-year TCO comparison.

Where inside the band do you land?

A band is a range, and four things decide where in that range you sit:

• Process clarity. A company whose process is written down, with known exceptions, lands near the lower end. Where “it depends” answers multiply, analysis time stretches.
• Data condition. Clean, single-source data shortens migration; multiple spreadsheets, duplicate records and hand-corrected rows extend it. To measure that separately, see the data migration estimator.
• Integration depth. Connecting through a modern API is not the same job as building file-based transfers. The integration complexity score measures that difference on its own.
• Decision speed. The most common cause of delay is organisational rather than technical: screen approvals waiting for weeks. Naming a single decision-maker shortens delivery noticeably.

Why starting small costs less

Looking at the list and ticking everything is a natural instinct — you do need all of it eventually. But taking everything in the first release inflates three things at once: the budget, the timeline and the risk of designing the wrong thing. A module nobody has used yet looks perfect before it is written; after a week in the field it becomes obvious which fields were unnecessary and which screens were missing.

So the sequence we recommend is: take the single process consuming the most hours, get it working in 2-4 weeks, let the team use it, then add the second module. The architecture is built for additions from the start. Your first invoice stays small, your first feedback arrives early, and the design of the second module rests on real usage rather than a guess.

If you cannot decide which process to start with, the criterion is simple: the one that repeats most often and produces the most errors. Order intake, stock counting and approval chasing usually fit that description. To find out which is most expensive, run the manual work cost calculator once for each of them.

What sits outside the band

The band on screen covers the build fee. These items are separate and all of them are published: hosting from year two onwards, optional maintenance at 12%, 18% or 25% per year, and modules added later. Declining maintenance does not stop the system; what stops is updates and new feature work. If you host the system yourself there is no hosting fee at all. The full list is on the pricing page.

Frequently Asked Questions