Blog

How long does it take to build custom software?

Scope sets the clock, slow decisions stretch it, phasing gets value out sooner, and a few weeks of discovery saves months of rework.

How long custom software takes depends far more on scope than on coding speed. A focused tool that does one job is a matter of months; a platform that runs a whole operation, such as a hotel property management system, a school system or an ERP, is a much longer build. The most useful thing you can do for your timeline is to be honest about which of those you are actually asking for.

This guide covers how project size sets the clock, what quietly slows a build down, how phased delivery compares with a single big launch, and why a proper discovery phase often shortens the total calendar instead of extending it. One caution first: the "typical timeline" figures that circulate online mostly come from development agencies' own marketing, not from studies, so treat any single number with care, ours included. What the research does show clearly is that size and decision-making drive the outcome.

Six people seen from above around a wooden table with laptops, a tablet and notebooks, two of them shaking hands across it

Project size sets the clock

The clearest way to think about a timeline is to size the project first. The categories below assume a competent team working steadily, with someone on the client side available to answer questions and make decisions.

  • Small: a single-purpose tool. One core workflow, a handful of screens, one or two integrations. A booking widget, an internal stock tracker, or a simple contact list for one team.
  • Medium: a departmental system. Several connected workflows, role-based access, reporting and a few external integrations. A clinic's scheduling and records tool, or a tour operator's booking back office.
  • Large: a platform that runs the operation. A hotel PMS, a school management system or an ERP touches many departments, carries complex rules, and often replaces several older tools at once.

Size also changes the odds. The Standish Group's 2020 CHAOS data, as summarised by OpenCommons (the report itself is paid), puts the success rate of small projects at about 90% and of large ones below 10%. The same feature list can land at very different points depending on data complexity, the number of integrations, and how many people must sign off on each decision.

What slows a build down

Engineering difficulty is rarely what blows a timeline. In research with the University of Oxford on more than 5,400 IT projects, McKinsey found in 2012 that large projects, those budgeted above $15 million, ran on average 45% over budget and 7% over time while delivering 56% less value than predicted. In its surveys of IT executives, weak management of strategy and stakeholders and weak command of the technology and project content caused about half of the cost overruns; team incentives and core project practices, such as short delivery cycles and rigorous quality checks, accounted for another 40%.

In practice, the delays look like this:

  • Shifting requirements. When the team is still discovering what to build while building it, finished work gets thrown away.
  • Slow decisions. A build waits on the people who approve designs, data and priorities. If a decision takes two weeks, the timeline absorbs those two weeks.
  • Integration surprises. Connecting to a payment provider, an old database or a third-party API often reveals behaviour nobody documented.
  • Unclear ownership. When no single person on the client side owns the product decisions, conflicting feedback stretches every review.
  • Scope creep. Each "small addition" mid-build carries a hidden cost in testing and coordination that the original plan never held.

The overall record shows how often this goes wrong: in the same CHAOS 2020 summary, 31% of software projects succeeded on time, on budget and with full scope, 50% were challenged and 19% failed.

Phased delivery vs big-bang launches

A big-bang launch holds everything back until one date, when the whole system goes live at once. Phased delivery ships working increments: the most important workflow first, then the next, until the system is complete. One of the principles behind the Agile Manifesto is exactly this: deliver working software frequently, every few weeks to a couple of months, with a preference for the shorter timescale.

Phasing rarely makes the total build shorter. What it changes is when value arrives and how much risk sits on any single day:

  • Value arrives earlier. A hotel front desk can be taking reservations in the new system while housekeeping and billing are still being built. A big-bang launch delivers nothing usable until the end.
  • Risk is spread out. A single launch date concentrates every unknown into one high-pressure event. Phasing surfaces problems a few at a time, while they are cheaper to fix.
  • Feedback shapes the rest of the build. Real users of an early phase show you what to adjust before later phases are built on a wrong assumption.

Big-bang launches still make sense in some cases, such as a hard regulatory cut-over or a system that cannot run half-migrated. For most operational software, phasing gives the company working tools sooner and a safer path to the full system. For how to choose the first phase, see how to build an MVP for your business software.

How discovery shortens the overall timeline

Discovery is the phase, usually a few weeks, where the team maps workflows, defines the data, lists integrations and agrees what the first release must do. It feels like a delay because no production code ships. In practice it is the most reliable way to protect the schedule.

Every decision made up front, such as how a reservation moves through its states, which fields a student record needs or how the billing rules actually work, is one the team does not have to stop and renegotiate halfway through the build. A good discovery phase produces three things: a prioritised feature list, a clear data model and a map of every external integration. With those in hand, estimates get tighter and the build runs with fewer interruptions. Skipping discovery to "start faster" usually trades a few weeks at the front for far more churn later. For more on this step, see how to scope requirements for a custom software project.

Setting realistic milestones

Milestones keep a timeline measurable, but only when they are tied to working software rather than paperwork. "Requirements document approved" tells you a document exists. "Reservations live for the front desk" tells you the product works. Set milestones a non-technical stakeholder can check by using the product:

  1. End of discovery: a prioritised backlog, a data model and an integration map, agreed in writing.
  2. First working increment: the single most important workflow running end to end with real data.
  3. Each later phase: a named module that users can log into and operate.
  4. Pre-launch: the full system running against production-like data, with the team testing real scenarios.
  5. Launch and stabilisation: go-live plus a defined window for fixing what real use exposes.

Build a buffer into the plan, sized to the unknowns discovery turned up: integration surprises, review cycles and the fixes that follow early feedback. A timeline with no buffer is a forecast that assumes nothing will go wrong, and something always does.

The fastest build is the one you do not need

Before commissioning a build, check whether one is needed at all. Setting up one of our ready-made systems, for hotels, sales, schools, operations and finance, clinics, tour operators, appointment bookings or professional services, takes days rather than months: we create the workspace, set up your properties, classes, products or pipelines, add your users and bring across your existing data, with no separate setup fee. When a business needs something one of them does not do, some requests are small enough that we simply add them, and larger ones become a custom module built on top of the same system. If one of our systems already covers 80% of what you need, we say so rather than quote a custom build. When a custom build is the right answer, it is quoted per project, in writing, with a timeline, before any work starts (pricing).

Questions about custom software timelines

How long does it take to build custom software?

It depends mainly on scope. A single-purpose tool is a matter of months; a platform that runs a whole operation takes much longer. Ask for a written timeline for the first working release alone, and treat a plan that cannot name one as a scoping problem.

What is the biggest cause of software project delays?

Unclear or shifting requirements and slow decisions. When the team is still working out what to build while building it, rework piles up and the timeline stretches well beyond the original estimate.

Is phased delivery faster than a single big launch?

Rarely in total build time, but it delivers usable software sooner and lowers risk. Users get value from the first phase instead of waiting for one large release.

Does a discovery phase make the project take longer?

Usually the opposite. A few weeks of discovery settles the questions that otherwise cause expensive rework mid-build, so it tends to shorten the overall calendar.

If you are scoping a system now, write down which size it really is and the single most valuable first workflow. That one decision does more to set a realistic timeline than any estimate a vendor can hand you. Tell us both and we will tell you whether one of our systems already runs it, or send a written plan and timeline for building it.