Stop trying to plant magic beans

If your “connected data” isn’t connecting much, maybe the “I” in your AI is missing. Here’s a reminder that information has to be intelligent before your tools can be, with some practical ways to rebuild your IM for the AI era. There’s a quick pulse at the end to test your own maturity.

When information was an art form

Do you remember when intranets were the new frontier and information management, portal strategies, taxonomy projects and governance frameworks weren’t just buzzwords?

We spent months designing navigation trees and arguing about taxonomy. Should it be People or Teams? Should you structure your information based on how the company is structured or what you do? Policy or Guidelines? Those were the days. We built labels, trained content owners and wrote governance documents that nobody read. It was fussy, detailed, often boring, but recognised as being essential.

Why did we and why did we want you to care? Because bad information is the reason CFOs go to prison. Or worse, your company ends up on the front page of the Daily Mail.

Before you scroll on, ask yourself how confident you are that your information management would pass a quick health check. There’s a five-minute test at the end if you’re brave enough to find out.

Because try as we might, it was mostly the big organisations that took information management seriously, mainly because chaos at scale is expensive. Smaller businesses wanted to skip that step, assuming proximity would keep things under control.

Ultimately governance is governance. The same risks apply whether you have fifty people or five thousand. One wrong permission or mis-filed document can still leak sensitive information or steer decisions off course. So we created lean versions, giving them just enough control to keep order without killing momentum.

Then the world moved on

Two big changes killed the craft.

First: “Digital workplace” thinking replaced “intranet” as the buzzword. The focus moved to experience and collaboration (Yammer, Teams, Slack) rather than structure and stewardship.

Second: cloud migration and Microsoft 365 arrived in force. Suddenly content lived everywhere: OneDrive, Teams, SharePoint, Planner, Loop. Plus local drives, shared drives that never quite got migrated and so on.

Every new Team created its own SharePoint site. Every channel spun up another folder structure. Before long, organisations had thousands of mini-intranets, each with their own logic, permissions and filing habits. Each with critical data only visible to a few.

Mini business-critical “applications” that IT didn’t know about, because Microsoft made it feel like DIY. Architecture exploded into tiny pieces.

Information management suddenly became bureaucratic and slow while the business wanted agility and self-service.

Microsoft gave us the power to classify, tag and manage content better than ever, but also made it easier than ever to ignore it. Feeding right into the CFO’s hands: stop worrying about structure, just get on and work. So they did. They created Teams for every project, sub-foldered to the moon and quietly broke the link between information and governance.

It was smart. It was momentum. Businesses needed speed to cope with the sheer volume of data compounding year on year. Nobody had the appetite to debate naming conventions or lifecycle rules. And the hard, unglamorous craft of information management faded into the background, replaced by a promise that the technology would somehow take care of itself.

From information to “insight”

Then a new game came to town. We fell for “data driven” and got peppered with shiny dashboards.

Data management, IM’s sexier cousin, handles quality, accuracy and storage. Information management is the layer above it: the human one that decides what’s worth keeping, what it means, who owns it and how it’s labelled so others can find it.

Without both layers, trying to get insight out of your information is like trying to write a novel with the library catalogue on fire. You can’t have one without thinking about the other. If you want any AI strategy to work, you need both.

Enter AI: the new miracle cure

The promise this time is that it will do the thinking for you.

• Plug it into your systems and it will magically tell you what’s going on in your business.
• Connect “all your data sources” and insight will fall from the sky. Or so it says.

Except, of course, it can’t. These tools can read words, scrape content and summarise with flair. But they can’t tell which version of Strategy_FINAL_v4_approved2 is the one that actually counts.

SharePoint indexes that field. It’s one of the main reasons people say the search “is rubbish”. I’ve lost count of the times I’ve seen Barry’s canteen menu as the title of a policy or board paper. A Word file copied, renamed, reused because it happens to have the right logo at the top.

AI can’t tell the difference between that and the real thing. It doesn’t know whether the sales report was built by an intern last week or signed off by finance. It can’t tell which SharePoint site your team actually works from or who’s responsible for keeping the template up to date.

AI is clever, but it’s still standing on the shoulders of your information management.

The inconvenient truth is that most businesses don’t have data science problems. They have information chaos.

Yes, some tools can trawl through the mess and produce something that looks coherent. They can even summarise conflicting documents and spit out a best guess. AI doesn’t know what’s current, credible or confidential unless you’ve told it. But it’s still guessing and you’re just automating ambiguity. That’s not intelligence. That’s inference.

IM isn’t dead. It’s the foundation

The old disciplines we quietly abandoned like taxonomy, metadata and stewardship are now the very things that make AI effective. If you can’t articulate how your business structures knowledge, AI can’t either.

So, sorry not sorry, we need to care about information management again. Not as a dusty governance exercise but as the foundation of modern intelligence.

Three simple truths

1. If your people can’t navigate your systems, neither can your AI.
2. If your teams don’t know what’s authoritative, your AI will quote the wrong thing with confidence.
3. If your information lives in silos, your insights will too.

Practical ways to rebuild IM for the AI era

1. Start where the pain shows. Identify the moments people complain they can’t find anything or don’t know which version is right. Fix those IM gaps before you buy another AI tool.
2. Don’t be afraid to give ownership back. Every major information set needs a named owner. Not IT or HR, not “the business”. An actual person. Accountability beats architecture.
3. Teach context, not systems. Your people think about information, not data, so start explaining why naming, tagging and archiving matter. It’s not admin designed to slow you down. Done sparingly and well, it’s intelligence.
4. Review your visibility map. I can’t stress this enough: make sure the right people can see the right things. That doesn’t mean locking everything down or working on a need-to-know basis. It means being deliberate about access. AI models draw from whatever they can reach, so don’t feed them blind.
5. Reward clarity. Celebrate the teams who label, tidy and document. They’re not pedants. They’re your competitive edge.
6. Stop planting magic beans. No good result ever came from no effort. The same is true of Excel!

We used to care about information management because it made work easier. Safer. Now we need to care because it’s the only way any of this really clever stuff will actually work.