← All essays

People & Systems · November 19, 2026

Owning Your Data Is the New Having a Website

Business owners who control and organize their data can use AI to improve decisions, strengthen workflows, and create new sources of value.

Thirty years ago, business owners wondered why they needed a website. I think we're having the same conversation again.

"What do you mean we need a website? We've got a brochure."

I remember those conversations.

It was 1997, 1998, somewhere around there, and I was involved with a number of businesses trying to figure out how to connect to this thing called the internet.

A bunch of us young guys would spend our evenings pulling blue Cat-5 cable through offices, connecting routers and switches, and helping companies get online.

It was exciting because you could see that something important was happening.

But most business owners didn't really understand why they needed any of it.

They already had brochures. They had telephones. They had salespeople. They had customers.

Why did they need a website?

Nearly thirty years later, I feel like I'm having the same conversations again.

Only now the question is different.

"What do you mean I need to own my own data? I'm already paying software companies to manage it for me."

I think we're at another one of those moments.

And I think the decisions business owners make over the next few years about their information could become some of the most important decisions they make.

Why I Decided to Bring Our Data In-House

Several years ago, I made the decision that we were going to bring our construction company's proprietary information into our own systems.

I wanted us to control it.

Not because I necessarily thought we could build better accounting software or project management software than everyone else.

But because I wanted to understand exactly what was happening inside our business.

I wanted our information organized in a way that made sense to us, rather than having to adapt our business to the way someone else's software works.

We've made considerable progress.

Today, most of the information associated with running our business lives inside systems we've built ourselves.

We're preparing to move payroll into our system on January 1, and our accounting application is the last major external system we're working toward replacing.

It's taken time. It's taken money. And it's taken a lot of work.

But I'm beginning to see how important that decision might have been.

Because now that artificial intelligence has become so accessible, the information we've been collecting is turning into something far more useful than a collection of records.

What Two Years of Data Has Taught Us

We've got almost two years of detailed operating information.

And we're starting to use it in ways that would have been much more difficult just a few years ago.

Take estimating, for example.

When we prepare a proposal, we can go back through our history and examine similar work we've completed.

How long did that particular task take?

What did we estimate it would cost?

What margin were we expecting?

And what did we actually earn?

If we estimated something at a 30% margin and the job came in at 20%, there's probably something we should learn from that.

AI helps us look through the information and identify those opportunities.

We're doing similar things with safety.

Instead of having our safety personnel prepare every piece of documentation from scratch, we can use AI to create a first draft based on the project.

Our people still review it and make sure it's appropriate for the work.

But they're not starting with a blank page every time.

We use AI for pre-job assessments, looking at scopes and schedules to help identify potential problems before work begins.

We've also developed an internal safety chat that helps employees find information from our safe work procedures, safety manual, and applicable regulations.

It doesn't replace competent supervision or the responsibility to verify safe procedures.

But it makes information easier to find.

And when you've got people working on different job sites, anything that makes the right information more accessible can be useful.

What If We Made a Thousand Small Improvements?

I think this is where AI becomes particularly interesting.

None of those things necessarily creates a sale.

Preparing a safety document faster doesn't put another dollar into revenue.

Neither does finding an answer in a manual or comparing one estimate against another.

At least not directly.

But each improvement saves time, reduces the possibility of mistakes, or helps someone make a better decision.

And those things compound.

We've all heard of death by a thousand cuts.

Businesses rarely become unprofitable because of one spectacular mistake.

Sometimes it's hundreds of little things.

A few hours wasted here. An estimate that's slightly wrong there. A delayed invoice. An employee waiting for information. A process that requires three people when it could require one.

Individually, they might not seem particularly important.

But together, they can make running a company extraordinarily difficult.

Now imagine going in the other direction.

What if we made a thousand small improvements instead?

What if the business became a little more efficient every week?

What if we made it easier for employees to do their jobs?

What if we made fewer mistakes and had better information when making decisions?

Over time, I think you end up with a very different business.

AI Might Be the Great Leveler

I spend a lot of time talking with entrepreneurs and business owners.

And one thing I've noticed over the years is that most business owners are really good at whatever it is their company sells.

That's usually why they started the business.

They're good at construction, manufacturing, plumbing, accounting, consulting, or whatever their particular product or service happens to be.

But running a business requires a completely different set of skills.

You need to understand finance, accounting, payroll, sales, management, human resources, systems, compliance, and countless other things.

And most business owners aren't particularly good at all of them.

Why would they be?

You don't become a great electrician by learning how to manage cash flow.

Yet the moment you start an electrical business, you're expected to understand both.

I was talking with my wife about this recently, and I said that it's such an interesting time to be running a business.

Because I think AI could make many of these responsibilities considerably easier.

A business might ultimately fail because it runs out of money.

But running out of money is often the consequence of other problems that weren't addressed early enough.

And those are exactly the kinds of problems I think AI could help owners identify.

It doesn't guarantee success.

You still need customers, good judgment, a useful product, and the discipline to execute.

But having access to tools that can help you understand the parts of business you don't know particularly well is a pretty remarkable opportunity.

Your Workflow Might Be More Valuable Than Your Software

A friend sent me a message this morning.

He'd been spending some time experimenting with AI development tools and was amazed by how easily he could create applications specifically for his business.

He made the observation that traditional enterprise software companies should be paying attention.

I think he's right.

For years, businesses have purchased applications for every particular activity.

We've got software for customer relationships, software for accounting, software for project management, software for time tracking, and software for just about everything else.

And every application has its own system for collecting and organizing information.

But what if the more important thing is understanding the workflow itself?

How does a lead become an opportunity?

How does an opportunity become a proposal?

How does that proposal become a job?

How does an employee's time become a payroll record?

How does completed work become an invoice?

If you can document those relationships and give AI enough context to understand them, all sorts of possibilities begin to open up.

It can help move information, prepare documents, identify missing steps, and support the people responsible for getting things done.

You don't necessarily have to buy another piece of software for every problem.

You might be able to solve some of those problems with the information and systems you already have.

That's the exercise I've been spending so much time on lately.

Your Data Might Become a Profit Centre

There's another part of this that I think could become very interesting.

What if the information a business produces becomes valuable beyond the products and services it sells?

Our company performs commercial demolition and asbestos abatement.

Over time, we've collected information about the work we estimate, how long tasks take, how projects perform, and what things actually cost.

That's specialized information.

And it's not necessarily the kind of information you'd find by reading a book or searching the internet.

I think there's a possibility that AI companies and other businesses will eventually be willing to pay for access to some of that knowledge.

Not customer names, private employee records, or information that we don't have the right to share.

I'm talking about properly structured, appropriately anonymized information that we have the right to license.

I don't know exactly what that market will look like.

But imagine a business that earns $1 million in annual profit.

Then the market becomes more competitive, margins shrink, and it only earns $200,000 from its regular operations.

What if that same business could license specialized information for another $500,000 or $600,000?

That would dramatically change its economics.

Those are hypothetical numbers. I'm not suggesting there's already a buyer waiting to write those cheques.

But I think it's an opportunity worth exploring.

Because it could mean that the information we collect while running a business becomes a source of value in its own right.

And if the licensing opportunity never develops?

Well, the information should still help us run our own business better.

That's a pretty good reason to organize it in the first place.

Anthropic Went Public Today

Anthropic, the company behind Claude, went public today.

I decided not to participate in the offering.

Not because I don't believe in AI.

I think AI is going to be incredibly important.

But I'm not sure I understand what the economics of these companies will look like twenty years from now.

I wonder whether AI eventually becomes something like the calculator on my iPhone.

I use it all the time.

But I don't think about who made it or what it costs.

It's just there.

Maybe AI becomes similar.

Maybe a handful of companies become enormously valuable.

I don't know.

And when I invest, I generally prefer owning businesses I can imagine holding for decades.

I haven't reached that level of conviction about Anthropic.

But the IPO has reinforced something I've been thinking about.

Regardless of which AI companies succeed, they're going to need useful information to make their products valuable.

And companies like ours produce specialized information every day.

That's the part of the opportunity that particularly interests me.

You Don't Necessarily Need to Build Your Own Software

I don't think every business needs to do what we've done.

Building your own systems is expensive and creates its own responsibilities.

And plenty of companies can operate perfectly well using existing software.

But I do think every owner needs to start thinking seriously about their information.

Where does it live?

Can you access it?

Can you export it?

Can you connect it to other systems?

Can AI understand what it means?

Are you allowed to use it for other purposes?

And if you decided to change software providers tomorrow, how much of that information could you take with you?

Using a software provider doesn't automatically mean you've surrendered legal ownership of your data.

But there's a big difference between legally owning information and having practical control over it.

That's the distinction I think many owners need to understand.

We've Been Here Before

Thirty years ago, business owners were asking why they needed a website when they already had a brochure.

Some jumped in.

Others waited.

And a few found ways to operate without one.

I think we're at a similar moment today.

Only now we're asking whether we really need to understand and control our own data when we've already got software companies managing everything for us.

My answer is that I think we do.

Not because every prediction about AI will come true.

Not because every business will become a data company.

But because I think the information a business produces is becoming increasingly important to its future.

And if you're planning to own your company for another ten or twenty years, that seems like something worth paying attention to.

I'm going to keep working through this in our own business.

I'm going to keep exploring whether we can turn some of that information into another revenue stream.

And I'll share what I learn along the way.

Because it feels a lot like 1998 again.

And I'm excited to see what happens on the other side.

Watch the episode

YouTube video thumbnailWatch on YouTube

Listen to the podcast