Gary may have been here 15 years but just accepting that “he knows things” is speedrunning catastrophe.
I once worked with Gary. He’d arrive every morning 8:56am on the dot clutching his briefcase containing his violently yellow coronation chicken sandwich, an apple and a satsuma (he corrected me once when I called it a nectarine) which he’d plonk on the left of the middle refrigerator shelf.
After creating his hot caffeinated liquid brown, he’d lift his laptop lid, boot up the company intranet and start copying. Manually. From a database that is older than Gen Z. Number by number into a spreadsheet Big G (to his mates) had built himself.
I once asked how he knew where the numbers were meant to go, the logic was as unfathomable to me as the opposite gender. “I am not being funny, I have been here 15 years, give me a customer code and I know where to add them”. That’s not written down, G Dawg isn’t going anywhere, he bleeds the company colours.
By 9:15am reports plop into relevant inboxes and the Grand Fromages make their decisions based on that. It works, it’s fine.
Why “It’s Fine” is a Misnomer in Governance
Reading that back, how much of that was automatic? Nada. So how much of it can go wrong? Everything else. Gary was the judge, jury and executioner of the daily reporting.
He’s a top bloke, but that doesn’t mean he won’t accidentally slip a 0 here or there. Or, accidentally paste a space into a cell which looks fine but frazzles formulas downstream. “02-11-2025”, bummer, I’ve opened the spreadsheet in a browser so is that November or February? On their own, nothing major causes alarm, but over the course of a year it might be putting the gross in Gross Margin.
Gary is also 58; he isn’t going to be here forever. And when he does chomp into the encased nuclear poultry for the final time, all his knowledge is going with him. What happens then? How is the 9:15am email going to happen?
The Gary 2.0
I could put this in terms of ETL, data warehouses and governed schemas, but even I would be bored before the end of this post.
Let’s instead run it back in an alternate universe. Gary goes and does his little morning ritual, lifts the lid of the laptop and bang, the report is there. The numbers have been pumped straight from the source system overnight, automagically, without a human’s fingerprint on them. The logic which was exclusively in his noggin has been converted to code, it’s written down, version-controlled and reviewable by anyone. When you pull “I think this milk is off” face, you trace the digits back to conception.
Sadly, there’s no more 9:15am spreadsheet emailed to 17 different people to be interpreted in 17 different ways to suit the receiver. Now it’s a singular version, all within the same reporting suite with predefined metrics and graphs that can be drilled into. The viewer can slice and dice it but only in ways the business has pre-agreed on.
It’s why I secretly get excited by a properly structured database with a curated reporting layer on top. Not just because I am a nerdy man, it’s because I know that these things empower a business to see the truth and make decisions based on reality.
8 Symptoms You Have A Governance Issue
- One person is the single point of failure for a business-critical report
- The logic behind your numbers exists in someone’s memory, not in documentation anyone else can read
- Data is copied by hand between systems at any point in the process
- You’ve had a ‘why don’t these figures match?’ conversation in a board meeting
- Reports take hours to produce and can’t be filtered, drilled into, or interrogated
- Nobody is quite sure what happens to the reporting function if that person leaves
- The spreadsheet has tabs labelled FINAL, FINAL v2, and FINAL USE THIS ONE
- When mistakes are found fixes are “quick and dirty”
Why This is Still Relevant In 2026
We are living in a weird world where some people are creating whole apps just by asking a bot nicely and others are creating tabs of chaos. The gulf between those two groups has never been wider, and it’s widening fast.
Businesses don’t find themselves here because of negligence. It’s because Gary never complained and the solution worked just about well enough to never be fully compelled to do anything about it. “Just about well enough” is a euphemism for “nobody realises how systemic the problem is, and nobody knows what good looks like”.
I fully understand that it may feel like you need to wear a robe, know the special handshake and speak in tongues to get involved with data. The language can get heavy quickly, and more often than not the people preaching aren’t the best at explaining it. It’s also not costing you anything extra for Gary to handle the numbers, as his salary is already baked in. You’ve probably seen the ‘me in 2016 vs me in 2026’ posts showing glow-ups, but there’s no glow-up here, this is definitely something we should be leaving in that time period.
The cost of something going wrong and the benefits of the validated, automated process heavily outweigh the work needed to do it. If not for you, do it for Gary.
