| Document Title |
|---|
| Excel is hard to quit: why it persists, where it becomes dangerous, and why AI makes data governance urgent | |
|
|---|
| Excel remains a universal tool for quick analysis, but becomes risky when it runs operations. AI adoption raises the penalty of messy, decentralised spreadsheet data. | |
| Title Attribute |
|---|
| oEmbed (JSON) | |
| oEmbed (XML) | |
| JSON | |
| View all posts by Admin | |
| Why more CEOs are sharing the top job: the case for (and against) co-CEOs | |
| Liquid cooling is becoming the bottleneck tech for AI data centres | |
| Page Content |
|---|
| Excel is hard to quit: why it persists, where it becomes dangerous, and why AI makes data governance urgent | |
| Nature | |
| Climate | |
| Why Excel won’t die: network effects, governance gaps, and the AI-era spreadsheet problem | |
| / | |
| Technology | |
| / By | |
| Admin | |
| Summary: | |
| Excel is 40 years old and still everywhere—even as organisations talk about modern data platforms and AI. The reason isn’t that Excel is “best practice.” It’s that Excel is a universal interface: flexible, teachable, and fast for small analyses. The danger is when spreadsheets quietly become production systems—undocumented macros, fragile workflows, and critical decisions built on files that aren’t centrally governed. | |
| The Excel story is really a story about how organisations manage (or fail to manage) data. | |
| Why Excel refuses to die | |
| From the BBC report: | |
| Excel remains widely used and embedded in education. | |
| It’s extremely good for quick analysis and charts on small datasets. | |
| Many organisations blur the line between analysis (fine in Excel) and processing/operations (risky in Excel). | |
| Excel succeeds because it is: | |
| low friction | |
| expressive | |
| locally controlled | |
| Those are user advantages—but governance disadvantages. | |
| The spreadsheet trap: analysis becomes infrastructure | |
| The BBC quotes an academic who describes departments where: | |
| data flows into spreadsheets | |
| macros transform it | |
| outputs feed important operations | |
| The risk: | |
| the macro author leaves | |
| nobody understands the workflow | |
| errors accumulate invisibly | |
| This is how “temporary” spreadsheets become permanent systems. | |
| Why AI makes the problem sharper | |
| AI is hungry for: | |
| clean, standardised, centrally accessible data | |
| Spreadsheets tend to produce: | |
| duplicated datasets | |
| conflicting versions | |
| unclear provenance | |
| local “shadow IT” processes | |
| So organisations trying to adopt AI often hit a wall: | |
| their data is trapped in people’s Excel files | |
| In that sense, Excel isn’t blocking AI because it’s old—it’s blocking AI because it decentralises data governance. | |
| The organisational reality: people want control | |
| A key insight in the BBC report is cultural: | |
| teams want to keep their Excel workflows | |
| they want new systems to export into spreadsheets | |
| This is understandable: | |
| Excel feels like ownership | |
| new systems feel like loss of control | |
| But for leaders, the data belongs to the organisation, not to individual files. | |
| Why replacing Excel is hard | |
| Excel is a general-purpose tool. | |
| Replacing it requires either: | |
| a suite of tools, or | |
| custom systems tuned to each workflow | |
| That’s expensive and disruptive. | |
| A more realistic strategy is: | |
| allow Excel for analysis | |
| prohibit Excel as a system of record | |
| That line must be enforced, or it collapses. | |
| Practical alternatives (and what they really do) | |
| The BBC describes businesses moving to: | |
| planning systems | |
| case management tools | |
| accounting platforms that extract invoice data | |
| These systems provide: | |
| structured data models | |
| permissions | |
| audit trails | |
| automation | |
| They reduce the risks of: | |
| silent edits | |
| version chaos | |
| undocumented transformations | |
| The hidden cost of Excel: operational risk | |
| Excel failures aren’t hypothetical: | |
| modelling errors | |
| copy/paste mistakes | |
| outdated files | |
| When spreadsheets run operations, the risk becomes systemic. | |
| That is why some organisations eventually force change by: | |
| not allowing the spreadsheet to coexist with the new system | |
| It sounds harsh, but coexistence often means “nothing changes.” | |
| What to watch | |
| Shadow IT | |
| : whether teams keep building mission-critical spreadsheets. | |
| Data governance projects | |
| : centralising and standardising data. | |
| AI adoption | |
| : AI will amplify the penalty of messy data. | |
| Better tooling | |
| : systems that preserve Excel-like flexibility with real governance. | |
| Bottom line | |
| Excel persists because it’s genuinely useful. | |
| The real problem is not that people analyse data in Excel—it’s that organisations quietly run critical processes in Excel. | |
| If AI is the next wave, the organisations that win won’t be the ones with the fanciest models. They’ll be the ones that finally get their data out of fragile spreadsheets and into governed systems. | |
| Sources | |
| BBC News (Technology of Business): | |
| https://www.bbc.com/news/articles/cwyxkzjpp87o?at_medium=RSS&at_campaign=rss | |
| ← | |
| Previous Post | |
| Next Post | |
| → | |
| oEmbed (JSON) | |
| oEmbed (XML) | |
| JSON | |
| View all posts by Admin | |
| Why more CEOs are sharing the top job: the case for (and against) co-CEOs | |
| Liquid cooling is becoming the bottleneck tech for AI data centres | |
| Excel remains a universal tool for quick analysis, but becomes risky when it runs operations. AI adoption raises the penalty of messy, decentralised spreadsheet data. | |
| |