The big platform is losing to the system you could just build

For twenty years the deal has been the same. Building software around the way your business actually works was slow and expensive, so you bought a big platform instead and bent your business to fit it. The platform did 70% of what you needed. You paid consultants to customise the other 30%, paid per seat forever, and paid again every time an upgrade broke the customisations.

That deal made sense when the alternative was a two-year build. It makes a lot less sense now.

The proof is already on your desktop

Look at the tools nobody thought could be replaced.

Word, Excel and PowerPoint. They are not going anywhere tomorrow, and the replacements are not perfect. But a huge share of what people opened them for — drafting a proposal, cleaning up a spreadsheet, building a twenty-slide deck for Tuesday’s meeting — is now something you describe in a sentence and get back in a minute. The document is no longer the work. The decision behind it is.

Image editing. Removing a background, extending a photo, resizing artwork for six social channels, mocking up a product shot — that used to mean a Photoshop licence and someone who knew how to use it. Most of it is now a prompt. Designers have not gone away; the routine work they were stuck doing has.

These were the most entrenched, most polished, most heavily invested products in software. If the routine parts of those can be replaced, the routine parts of your CRM can too.

The big, expensive systems

The pattern is the same everywhere a platform is sold as “configure it to fit”.

Salesforce. A genuinely capable platform that most companies use as an expensive contact list with a pipeline view. The features you actually rely on are a small slice of what you are licensed for, and the slice that is specific to your business lives in custom objects, flows and Apex code that only one consultant fully understands.

Microsoft 365. Mail and calendar are a commodity and worth paying for. But the suite sells itself as the place your business processes should live, and the moment you try to make it do that — approvals, intake forms, a tracker that three departments share — you are stitching together Power Automate, Lists, Forms and Teams and hoping nothing gets renamed.

SharePoint. No one — not one person — has ever ended up liking SharePoint. It is bought as the answer to “where do our documents and processes live”, and it never quite does what you want. So you customise it. Web parts, workflows, permission schemes, a partner who built the intranet. Then Microsoft changes the platform underneath — classic to modern, the 2010 and 2013 workflows retired, InfoPath at end of life — and the custom work breaks. Every few years you pay to rebuild what you already paid to build, on a platform you never chose because you liked it.

That is the real cost of a big platform: not the licence, but the permanent tax of making someone else’s general-purpose product do your specific job.

What changed

Building software got dramatically cheaper.

AI did not make engineering free, and anyone telling you it did has not shipped anything to production. What it did was collapse the cost of the product layer — the screens, the forms, the reports, the business rules, the integrations, the tests. The parts that used to make a custom build take eighteen months now take a fraction of that.

Which flips the arithmetic. When building your own was a two-year, seven-figure project, renting a platform and customising it was the safe choice. When a system shaped exactly around your process costs less than a few years of licences and consultant hours, the platform is the expensive option — and the one that breaks on someone else’s schedule.

A system you own does exactly what your business does, and nothing else. No seats you never use, no modules you are paying for but ignoring, no upgrade cycle you do not control. When the process changes, the software changes with it, because it was written for you in the first place.

What didn’t change

Here is the part the hype leaves out.

You still need servers. Software has to run somewhere — hosted, backed up, monitored, secured and kept patched. Cancelling a SaaS subscription means taking on the job it was quietly doing for you. That job is smaller and cheaper than it used to be, but it is not zero, and it is not optional.

You still need your own data and your own processes. This is the actual asset. The platform was only ever a container for it. Moving off Salesforce or SharePoint is mostly a data problem: what you have, where it lives, how clean it is, what depends on it, and who is allowed to see it. Get that wrong and the shiny new system is just a faster way to show people bad data.

You still need someone to collect the requirements and build the system around your data. AI can write a convincing app from a one-paragraph description. It cannot sit with your operations lead and find out that the approval step everyone complains about exists because of a regulatory audit in 2019. It cannot tell which of the forty fields on your opportunity record anyone actually uses. It produces plausible software quickly; making it correct for your business is still the work of someone who asks the right questions and knows what breaks in production.

That is the real shift. The scarce skill is no longer typing the code. It is understanding the business well enough to know what to build, and engineering it well enough that it keeps working.

So should you rip out your platform?

Not on reflex. Mail, calendar and file storage are commodities — keep paying for them. But if any of these sound familiar, it is worth pricing the alternative:

• You pay for hundreds of seats on a platform most people use for one or two screens.
• A meaningful slice of your budget goes to a consultant whose job is keeping customisations alive through upgrades.
• Your team keeps spreadsheets alongside the system because the system will not do what they need.
• Nobody can change a workflow without raising a ticket with the one person who understands it.
• The platform’s last major version change forced a rebuild you did not plan for.

The question is no longer “can we afford to build our own?” It is “how much longer can we afford not to?”

That is the work we do: sit down with the people who run the process, work out what the data and the workflow actually are, and build — and host — a system around them that is yours. If you are paying a platform to be almost right, talk to us.