The most positive question: How will your tech upgrade fail?

Every week I sit across from a business owner who has spent serious money solving a problem in the worst possible way.

It goes like this, almost like clockwork: The system they have works, but they know they need an upgrade as their business has grown and/or wants to grow further. So they go looking, and they find a system that is newer, offers a better dashboard, and they were impressed after being walked through a compelling demo. So they buy it.

Six months later, those operations that used to be so simple to handle with the old spreadsheet are broken in this "advanced" system, and nobody can say exactly why.

It's a common story addressed with old wisdom that is worth reviving, especially now. In the nineteenth century, the mathematician Carl Gustav Jacobi kept a standing instruction for hard problems: invert, always invert. Turn the question around. Charlie Munger built an investing career on the same move, asking not how a decision would succeed but what would destroy it, then avoiding those things. The discipline works just as well for technology. Stop asking how a new system will help. Ask how it will fail, then stay clear of those failures.

This matters more in 2026 than it did five years ago, because the temptation to buy has never been stronger. AI has turned every vendor into a visionary and every board meeting into a race not to be left behind. Shiny object syndrome is at its sharpest right now, and it is expensive. Each switch carries a hidden bill: migration hours, re-learned workflows, and data scattered across platforms that were each sold as the single source of truth.

We recently pulled these failure patterns into a guide, The Big Five: What Not to Do When Implementing New Systems, which also sets out the ten questions every project team should answer before a contract is signed. Within it, I outline the failures I see the most consistently:

• The first is chasing features over fundamentals. One recent example I worked with had replaced one older, unglamorous platform with six best-in-class applications. On paper it was an upgrade. In practice, invoicing broke and loyal customers walked through the door to a system that had no idea who they were. The new stack was judged on what it added. Nobody confirmed what it preserved.

• The second is buying hardware when software already solves the problem. A manufacturer nearly committed to an expensive, hardware-based timekeeping rig, sold hard through their industry network. An established off-the-shelf tool did the same job, geofencing and all, for under $250 a month with nothing new to install. Most operational problems have already been solved by proven, unglamorous software. Bespoke builds belong at the end of the queue, once the cheap and boring options are exhausted.

• The third is researching to confirm rather than to test. Too many organisations research technology the way people read car reviews after they have bought the car. The decision is already made, and the research exists to justify it. This bites hardest at integrations. Every vendor says their product connects to Xero or QuickBooks. I have watched well-regarded tools perform beautifully at their core job, then fail completely at pushing clean data into the accounting file. Every integration claim is a hypothesis until you have tested it with your own data, live.

• The fourth is taking compliance claims on trust. In that same hospitality project, the supplier was new to the Australian market and assured everyone the GST configuration was handled. It was not, and nobody found the errors until they were live. Tax treatment is jurisdiction-specific, unglamorous, and invisible in a sales demo. World-class engineering is no guarantee that someone understands how GST applies across your revenue streams.

• The fifth, and the one most often sacrificed when timelines slip, is skipping user acceptance testing. The people who will live with a system every day, and who understand the compliance obligations best, are exactly the ones who need a seat at the table. Too often they never get one.

You might reasonably push back here. Surely a business that avoids every risk simply stagnates, and in a market moving this fast, standing still is its own kind of failure. That is fair, up to a point. The aim is to change for a genuine reason, an operational improvement for the people doing the work, rather than for a KPI, a committee, or the fear of missing out. Underneath all of it sits an uncomfortable truth: the highest-return technology investment in most businesses is training. Staff who click cancel on every update and who never learned the software already in front of them leave value on the table every single day.

None of this is an argument against new systems, and with the right, strategic approach that gets the balance right, good technology quietly compounds value for years. All you need to do is invert the question and ask how the solution could fail first.