How long does it take to build custom software for a small business?

By Ranjit Bhinge, founder of Blur Studio · Updated · 7 min read

Short answer: For one business process, such as a job tracker or a customer database, in my planning a first working version takes a few weeks, not months. Bigger systems with many connected parts take several months. Most delays come from the owner side: late decisions, messy data and slow approvals, not from typing code. A first version is the smallest piece of software that your team can use for real work.

Ranjit Bhinge

You want to know how many weeks or months pass between saying yes to a custom build and your team actually using it.

I'm Ranjit, and I run Blur Studio: 15+ years building software, 120 apps taken live, and clients who have gone on to raise $10M+. I build AI systems and processes that help businesses do a lot more with a lot less time. I also work differently from most of what else is out there, more on that further down this page. Say hello.

Quotes from large agencies often run to many months because they cover big projects with many teams. That does not fit a small business that wants one process fixed. The useful answer is a timeline by project type, plus a list of the things you control that decide whether it takes four weeks or four months. If you run a freight brokerage, see custom software for freight brokers.

Two meanings of "built"

People use the word to mean two things, and mixing them up is the main reason estimates disagree.

  • First usable version. The team can do real work in it, even if it is plain and missing features.
  • Finished system. Every feature you ever wanted, polished, connected to everything.

I plan for the first one. The second one never really ends, because your business keeps changing. A good plan gets you the first version fast and then improves it in small steps.

Timelines by project type

These ranges are my own planning judgment for a small project with one decision-maker on the owner side. They are not an industry average, and they assume the work is scoped to one process. Your project can land outside them. For the running cost after launch, see what software maintenance costs per year.

ProjectPlanning range for a first usable versionWhat drives it
Job and customer tracker2 to 5 weeksNumber of statuses, whether staff need phone access
Simple internal tool (approvals, request forms)2 to 6 weeksWho approves what, email alerts
Customer or client portal5 to 10 weeksLogins, permissions, what customers can see
CRM for a small team5 to 10 weeksData import, email and calendar links
Quoting and invoicing system6 to 12 weeksPricing rules, tax, accounting link
Custom ERP (one system for stock, orders, accounting)4 months or more, in stagesNumber of departments, existing data

For the first two rows, a worked example of scope is in how to build a simple job and customer tracker without a $10,000 quote. For the last row, see custom ERP, where staging is the whole point.

What public sources say

You do not have to take my word for the shape of this. Some public benchmarks:

  • The UK government's service manual says an early discovery phase, where a team learns the problem before building, typically takes about 4 to 8 weeks, and says there is no set period. It also says you should not start building during discovery.
  • The same manual says an alpha phase, where a team builds rough prototypes to test ideas, usually runs about 6 to 8 weeks, and that prototypes are expected to be thrown away.
  • The same manual says that when you replace an existing service, you keep the old one running until the new one goes live, which is the same idea as running your old way next to the new tool for a week.
  • Basecamp (a project management software company) publishes a method called Shape Up. It treats six weeks as the standard unit of build work and sizes small projects at one to two weeks for a small team. It sets a time budget first and then cuts scope to fit, instead of estimating and hoping.

Government services are bigger than a small business tool. I use these as a reminder that learning the problem, prototyping, and building are separate stages with separate lengths. A small project compresses them, but it does not remove them.

What really slows a project

In my experience the delay is rarely the building. It is waiting. These are the five that cost the most time, in rough order.

  1. Slow decisions. A builder asks "should a job be able to have two owners?" and the answer takes a week because three people have to agree. Name one decision-maker at the start.
  2. Messy data. Duplicates, missing phone numbers, five spellings of the same customer. Cleaning takes longer than importing. Start cleaning on day one, in parallel.
  3. Approvals and access. Waiting for logins to your email, accounting tool or domain. Gather these before the build starts.
  4. Scope growth. "While you are in there, can it also do payroll?" Each addition is small. Together they double the time. Keep a not-now list.
  5. No time to test. The builder finishes a screen and nobody looks at it for ten days. Block 30 minutes twice a week for feedback.

What you control and what you do not

You control the five items above. You do not control how long it takes a skilled person to build a given screen, and you should not try to squeeze that. If a builder promises half the timeline of every other quote, ask what they are leaving out. It is usually testing, data import or documentation.

The one-process-first approach

The fastest route I know is to pick the single process that hurts most, build only that, and put it in front of your team. For many small businesses that is the place where jobs, leads or orders get lost between a spreadsheet and an inbox. If you are not sure it is worth building at all, a minimum viable product (a first version made to test whether the idea works) is the right lens. I cover that in AI MVP development.

A rough rhythm I use for a small build:

  1. Week 1: map today's steps, agree the not-now list, collect data and logins.
  2. Weeks 2 to 3: build the core screens and a clean data import. You test every few days.
  3. Week 4: your team runs real work in it next to the old way for a week.
  4. After: fix what the week showed, then decide what version two should add.

Worked example, with made-up figures: a plumbing firm of 8 people tracks jobs in a spreadsheet and text messages. The owner names herself decision-maker, sends a cleaned list of 600 customers in week one, and gives 30 minutes twice a week for feedback. A first tracker, with job status, customer notes and a reminder for follow-ups, goes live in four weeks. The same firm, with no decision-maker and a spreadsheet nobody cleaned, could take three times as long for the same software. The code was never the difference.

How to read a timeline in a quote

When a developer gives you a date, ask what the date means. Four questions sort good quotes from vague ones.

  • What will I be able to do on that date? "Your team enters real jobs" is a clear answer. "Development complete" is not. GOV.UK's guidance on user stories says the goal is the most important part, because it helps you decide when the work is done.
  • When do I see something working? You should see a rough version within the first two weeks. If the first look is at the end, you carry all the risk.
  • What do you need from me, and by when? A good builder lists your tasks with dates: the data file, the logins, the decisions. If the timeline does not mention you at all, it assumes you never cause a delay, which is rarely true.
  • What happens if we add something? Ask how changes are handled. The answer should be that something of equal size moves to the not-now list, or that the date moves and you are told by how much.

The time budget idea from Shape Up helps here. If you decide six weeks is what the project is worth, the conversation becomes "what is the best version we can finish in six weeks," which is a better question than "how long will everything take."

When buying is faster

If a ready-made tool covers 90 percent of what you need, buying is faster than any build. You can be running in days. That is the point of build vs buy software. Custom wins when the missing 10 percent is the part that makes your business different.

How I'd approach this

I would set a time budget first. Four to six weeks for a first usable version of one process, and I cut features until it fits. That keeps the project honest and gives you something real to react to. If a quote says six months and you only asked for a tracker, push back on the scope, not the pace.

Before you contact anyone, do three things: name one decision-maker, start cleaning your data, and write the not-now list. Those three save more weeks than any developer can. If you want to talk it through, see what a build with me includes, and bring your process and your user count to a quick call. If you are weighing it against a cheaper route, read how to build an internal tool without a developer as well.

Questions people ask

How long does it take to build a custom CRM?

For a small team, I plan around 5 to 10 weeks for a first usable version, as my own planning judgment. Data import and email or calendar links are the usual time sinks. Cost is a separate question, covered in custom CRM cost.

How long does it take to build an MVP?

An MVP (minimum viable product) is the smallest version that tests whether your idea works. A narrow one can take a few weeks. See AI MVP development for how AI tools change that, and what still needs a person.

Why do agency quotes say 6 to 12 months?

Quotes in that range usually cover larger scope: many user types, many connections, formal testing and approvals. A single-process tool for a small team is a different, smaller project, so ask for a quote scoped to one process.

Can I get a working version in a few weeks?

Yes, if the scope is one process, you have one decision-maker, and your data and logins are ready. Plan a week of running the new tool next to the old way before you switch.

What delays a custom software project the most?

Waiting on the owner side: slow decisions, messy data, missing access and unreviewed work. Scope growth is the next biggest, so keep a not-now list.

Should I wait until requirements are perfect before starting?

No. A one-page brief is enough to start. If you want a template, see how to write requirements for a software developer.

More in this topic: Problem solving: custom software (22 guides) →

Sources

Ready for the next step?

A 30 minute call. If you do not need me, I will tell you.

Ranjit Bhinge You talk to me, Ranjit. I reply within a few hours.

Not sure what you need? Tell me what is going on. I will point you the right way, even if it is not me.

Blur Studio · 30 minute call

Book a 30 minute call

You talk to me, Ranjit. If you do not need me, I will tell you.

  • We talk about your business and what is going on.
  • You leave with a clear next step, even if it is not me.
  • I reply to every message within a few hours.
Loading the calendar…