How long will your software project take?

Eight short questions about what you want to build. The answer is a range, not a promise, shown with every assumption behind it.

Everything is worked out in your browser
About a minute
A range, not a price

How we work it out

Every number on this page comes from this list. There are no hidden ones.

Effort (person-weeks) = the base for the kind of product + the features chosen + connections to other systems + data migration.
From the fourth feature on, each one adds 2% to the effort (up to 20%), because features meet each other and bugs live in the seams.
The effort is then multiplied by (1 + the data-sensitivity uplift).
Pace: 2 engineers at 75% useful focus (the rest of the week is review, meetings and switching), which is 1.5 person-weeks of progress per calendar week. A job under 8 person-weeks cannot be split between two people and runs at one person’s pace (0.75).
Build time = effort ÷ pace. Fixing and acceptance add 20% of the build time.
Waiting for your feedback is a share of build and fixing time (table below).
On top come discovery, a part of the design (60% of the designer’s weeks land on the calendar; the rest overlaps with discovery and set-up) and launch (1 week, or 2 with a lot of data). The total is the middle estimate.
The range = the middle × a low factor and a high factor, which come from how clear the requirements are (table below). The high factor gains 0.1 for each risky answer, up to 0.3: an old system, a lot of data to move, formal requirements, slow feedback.

Base for the kind of product (person-weeks)

An internal tool for staff: 5
A public site your team can edit: 4
A web app for customers: 8
A mobile app and the API behind it: 14
A reporting dashboard over existing data: 4
Connecting systems you already run: 3

Each feature (person-weeks)

Roles and permissions: 1.5
Online payments: 2
SMS and email notifications: 1.5
Reports and Excel or PDF export: 2
File and media upload: 1
Search and filters over a lot of data: 1.5
An admin panel: 2
More than one language: 1.5
Workflow and multi-step approvals: 3
Real-time features: 3

Connections to other systems (person-weeks)

No, it stands alone: 0
One system, with a clear API and documentation: 2
One older or undocumented system: 4.5
Several systems, all with a clear API: 5
Several systems, some old or undocumented: 10

Data migration (person-weeks)

No, we start fresh: 0
Not much, and tidy: 1
A lot, or messy: 4

Design (designer-weeks)

The final design is ready: 0
Standard components are fine: 1
It needs a custom design: 3

Clarity of requirements: discovery weeks and the width of the range

Written down and specific: 1 week of discovery · Range 0.8× to 1.3×
A general idea, not the detail: 2 weeks of discovery · Range 0.65× to 1.6×
Just the problem: 3 weeks of discovery · Range 0.5× to 2×

Data sensitivity: uplift on effort

Ordinary: +0%
Customers’ personal or financial details: +15%
Formal requirements: +35%

Speed of feedback: uplift on build time

Within two working days: +0%
Within a week: +15%
Sometimes two weeks or more: +40%

Basis: a simple bottom-up model in the spirit of reference-class estimating, and the well-known idea that an early estimate is inherently wide and narrows as the requirements become clear. These figures are working assumptions, not measurements: they have not been fitted to any particular team’s records. Read them as a starting point for a conversation.