Build vs Buy? An Old Question With New Answers

I’m old enough to remember when you could walk into a computer shop and buy software in a box.

MYOB. Quicken. Antivirus software. Microsoft Office.

You bought a box, took it home, inserted a collection of increasingly annoying floppy disks—or, later, a CD-ROM—and installed your software.

Off-the-shelf software represented extraordinary value.

Highly skilled business analysts, software architects, engineers and developers could invest thousands of hours building a product, then recover that investment by selling copies to tens or hundreds of thousands of customers.

You didn’t get software built specifically for you.

But you received the benefit of an enormous amount of development work for a tiny fraction of what it would have cost to build yourself.

It was a pretty compelling equation.

Then the internet arrived.

At first, it mostly gave software companies a better way to distribute software, provide updates and verify licences.

Support contracts and software licensing also introduced the industry to something even more attractive than repeatedly selling software:

Recurring revenue.

And software companies rather liked recurring revenue.

As internet connections became faster, the application itself could move onto the internet. Nothing to install. Nothing to update. Log in from anywhere.

Enter SaaS: Software as a Service.

And what a service it has been—and still is.

The old "Build" vs "Buy" question

Throughout all of this, businesses have faced essentially the same question:

Should we buy something, or build something ourselves?

Sometimes “build” didn’t mean hiring software developers.

It meant the terrifying Microsoft Access database sitting on that computer in the back room.

Or an increasingly elaborate Excel spreadsheet that began life as a simple list and somehow became responsible for half the company’s operations.

Tools such as Access, FileMaker and Excel allowed businesses to build systems around the way they actually worked.

They were cheap. Sometimes they effectively appeared free because they came bundled with software the business already owned.

But purchase price and true cost are very different things.

That Access database still needed somebody who understood it.

The spreadsheet required three days of someone’s time every month.

When Susan left, everybody discovered that Susan was apparently the only human being on Earth who understood how FINAL_REPORT_v7_ACTUAL_FINAL_USE_THIS_ONE.xlsx worked.

So businesses bought software.

And usually that was the right decision.

Why "Buy" kept winning

Custom software has always been an option.

The problem was the price.

A serious custom business application traditionally required business analysts, architects, developers, designers, testers, project managers and infrastructure expertise.

Building something substantial could easily cost hundreds of thousands of dollars—sometimes millions.

Against that, paying a few thousand dollars each month for an established SaaS product made enormous sense.

Sure, the software wasn’t designed specifically for your business.

You might have to change some processes to suit it.

You might need another product to fill a gap.

Then another integration to make those products communicate.

Then perhaps some spreadsheets to deal with the bits neither product handled particularly well.

But compared with building your own software?

"Buy" won.

For most small and medium businesses, it wasn’t even particularly close.

But one of the fundamental parts of that calculation has changed.

AI has changed the economics of "Build"

AI has become exceptionally useful in software development.

It is not perfect.

It is not autonomous.

And it is certainly not a substitute for understanding the human and business problem the software is supposed to solve.

But modern AI tools can produce, analyse, test and modify code extraordinarily quickly. They can remove a substantial amount of repetitive development work and significantly increase what a small, experienced team can deliver.

Place that capability in the hands of someone who understands software architecture and can sit with a business and understand how its people actually work, and something interesting happens.

The cost of producing custom software comes down.

Way down!

In my own work, AI-assisted development has dramatically increased the amount of software I can produce as one person. Work that would previously have required several developers can increasingly be designed, built and tested with a much smaller human team directing and supervising AI driven development tools.

The exact productivity improvement will differ enormously between projects, developers and methodologies.

AI does not remove the need for architecture, security, testing, product judgement or accountability. Producing code quickly is useful. Producing the right system is what matters.

But the direction is clear:

The economics of producing software are changing.

And that means businesses should probably dust off the old "Build vs Buy" calculation.

Let’s do some deliberately simple math to illustrate this.

Suppose a business relies heavily on a SaaS platform costing $3,000 per month.

That is:

$36,000 per year.

Or:

$108,000 over three years.

Obviously, the SaaS provider is not just banking that money.

It is hosting the application, maintaining it, developing new features, fixing bugs, providing support, handling security and doing all sorts of things that must still be considered if you own the software yourself.

Custom software also has ongoing costs. It needs hosting, monitoring, maintenance, security updates and support. It may rely on external services with their own fees and licence conditions.

But here is the interesting question:

What could you own for $108,000?

Ten years ago, the answer might have been:

Not even nearly enough.

Today?

A lot

I think businesses should actually run the numbers.

Because $108,000 can now buy a surprisingly substantial amount of AI-assisted custom software development.

Depending on how the project is structured, and what it needs to achieve $108,000 could buy the business its own bespoke software and its intellectual property, increasing the value of the business!

Importantly, custom software is designed to fit your business.

Your terminology.

Your workflows.

Your customers.

Your strange exceptions.

Your competitive advantages.

The peculiar way you have discovered over twenty years that your business works best.

You do not necessarily have to change your business to fit the software.

The software can fit the business.

But replacing SaaS isn’t necessarily the answer

This is not an argument that every business should cancel its SaaS subscriptions and start building software.

That would be ridiculous.

Sometimes buying is overwhelmingly the right answer.

Microsoft 365 costs vastly less than building your own Word, Excel, email platform and video-conferencing system.

There are enormous advantages in buying mature software where thousands or millions of customers genuinely have essentially the same requirement.

But that is not every business problem. Ask yourself how many of the following examples sound familiar to you and your business?

Perhaps your expensive SaaS platform is actually excellent, except for one or two workflows that drive everybody insane.

Don’t replace it.

Fix the workflows.

Perhaps staff spend three days every month manually transferring, reconciling and checking information between two systems.

Connect the two systems.

Perhaps somebody maintains a monstrous spreadsheet because your core system does not quite do something the business needs.

Replace the spreadsheet.

Perhaps you are paying for four different SaaS products because each solves a different part of your unique business puzzle.

Now that is really worth investigating. Because integration or custom software might be a great fit.

Because the real cost of your software is not simply the number appearing on your credit-card statement every month.

It is something more like:

Subscription costs
+ implementation costs
+ integration costs
+ support costs
+ staff training
+ duplicated effort
+ manual workarounds
+ additional software required to fill the gaps
+ the cost of changing your business to fit somebody else’s software

That is the true cost of your system of work.

And that is the true "Buy" number that "Build" needs to compete against.

Run the numbers again

"Build vs Buy" is not a new question.

Businesses have been asking it for decades.

For most small and medium businesses, the answer increasingly became "Buy" because the economics of professionally developed custom software simply could not compete with software whose development costs were distributed across thousands of customers.

That logic was sound.

But the calculation contained an assumption:

Software is expensive to build.

AI is changing that assumption.

It does not mean "Build" suddenly wins every time.

It means the numbers have changed enough that the question is worth asking again.

So take one of those subscriptions you have been paying every month.

Multiply it by 36.

Then add the integrations, workarounds, spreadsheets, duplicated effort and human hours required to make everything work together.

Now ask yourself two questions:

What could we build and own for that money?

And perhaps more importantly:

How much less pain could we "buy" by "building" something that actually fits?

That is the question I’m helping businesses answer.