There’s a folder on your shared drive called “Misc.” Or maybe it’s “Old Stuff”. Perhaps it’s the optimistically named “To Sort Later” from 2019. We’ve all got one. And most of us have quietly agreed never to open it.
Here’s the thing: that folder isn’t just a mild embarrassment. If you’re planning to use AI tools in your business, that digital junk drawer could be the reason your shiny new technology delivers disappointing results.
The Old Computing Principle That Still Applies
“Garbage in, garbage out” has been around since the 1950s. It’s one of those phrases that sounds obvious until you realise how often we ignore it. AI tools work by processing your business data to find patterns, answer questions, and generate insights. When you ask an AI assistant to summarise your company’s approach to customer complaints, it searches through your files to piece together an answer. If those files include three different versions of your complaints policy (one from 2018, one draft that was never finished, and one with “FINAL_v2_ACTUALLY_FINAL” in the filename), the AI will dutifully try to make sense of all of them. The result then gives you confused, contradictory, or simply wrong outputs. And suddenly that expensive AI investment starts looking like a waste of money. Research from Experian found that while 90% of UK business leaders agree high-quality data is essential for AI, only 43% are confident their data is strong enough to support it. That’s a significant gap between knowing data matters and having data that’s fit for purpose.Common Data Hygiene Issues (And Why They Matter)
Before you dismiss this as someone else’s problem, let’s run through the usual problems. See how many sound familiar. Duplicate files everywhere. The same document is saved in three locations, with slight variations that nobody can remember making. When AI tools search your data, duplicates create conflicting information that muddles results. Lack of naming conventions One person saves files as “ClientName_Project_Date”. Another prefers “2024_January_Report_Draft”. Someone in accounts just calls everything “Document1”. AI tools rely heavily on file names and metadata to understand what things are. Inconsistent naming makes their job significantly harder. Outdated information living alongside current data. That one pricing document for a discontinued service is still sitting in the Sales folder. The AI doesn’t know it’s obsolete. It just sees another source of information to draw from. Folder structures that made sense once. Complex, nested folder hierarchies that require institutional knowledge to navigate aren’t just frustrating for your team. They’re genuinely problematic for AI tools trying to locate relevant information.The Permissions Problem
Here’s something that catches many businesses off guard. When you adopt AI tools, particularly those built into platforms like Microsoft 365, they inherit whatever access permissions your users already have. This sounds sensible until you realise what it means in practice. That accounts assistant who somehow has access to the HR folder containing salary information? An AI tool will happily surface that data if they ask the right question. The intern who was temporarily given access to client contracts and never had it revoked? Same story. Before AI, these permission gaps were theoretical risks. Someone would have to actively go looking for information they shouldn’t see. With AI tools, they just need to ask a question, and the system does the searching for them.Quick File Structure Fixes That Make a Big Difference
So, let’s talk about fixing this. The good news is you don’t need to achieve perfection. You just need something that’s good enough. Start with your most-used folders. Don’t try to reorganise everything at once. Identify the folders your team accesses daily and focus there first. Sales documents, client files, operational procedures. Get these in order, and you’ll see immediate benefits. Agree on naming conventions and write them down. It doesn’t matter hugely what system you choose, as long as everyone follows it. Date formats, version numbering, project codes. Document your approach and share it with the team. Archive rather than delete. Not sure if something’s still needed? Move it to a clearly labelled archive folder rather than leaving it mixed in with current files. This keeps your working folders clean without the anxiety of permanent deletion. Set a quarterly review schedule. Data hygiene isn’t a one-off project. Like cleaning, it’s a habit worth forming. Build in quarterly reviews to catch drift before it becomes chaos again. If the scale of the task feels overwhelming, or you’re not sure where your biggest data risks lie, an IT strategy session can help you prioritise what to tackle first.The Business Case for Getting Organised First
We understand the temptation to skip straight to the exciting AI implementation. But consider this: would you hire a brilliant new analyst and then hand them a filing cabinet stuffed with unsorted papers, obsolete reports, and documents nobody can identify? The AI tools available today are genuinely impressive. Microsoft Copilot, for instance, can draft emails, summarise lengthy documents and find information across your entire Microsoft 365 environment. But these capabilities depend entirely on having organised, accurate and properly permissioned data to work with. Getting your data house in order first means:- AI tools can find what they’re looking for
- Results are based on current, accurate information
- Sensitive data stays visible only to appropriate people
- Your team trusts the outputs (because they’re reliable)
- You get genuine return on your AI investment
Ready to Make AI Practical?
Join Cloud Geeni for a free Lunch & Learn session where we’ll cut through the hype and show you practical steps to make AI work for your business. Learn what good data looks like, how to avoid permission pitfalls, and quick wins that drive real value. Register your interest for the Making AI Practical event 