Why are organisations surprised when AI delivers no benefit? Because they are focusing on the 10% that is the tool.

Introducing AI is fundamentally different from earlier technology change. With ERP, CRM and SCM we automated existing processes. Sometimes we documented them first. Often we bent the process or the system to get a "successful" go-live. Either way, the work stayed broadly the same, only faster.

Despite the excitement around Gen/Agentic AI, most organisations are following that same well-trodden path and expecting the same kind of payback.

Microsoft #Copilot is the case in point. Every large Microsoft shop I talk to is rolling it out, based on well-meaning stories: automated meeting notes, meeting prep pulled from past contact, faster decks, help with that challenging Excel or Power BI model. 30mins here, 20 there,15 somewhere else, and it all adds up to a productivity case.

It doesn't. Despite the Copilot training, the time saved gets converted into more stuff (AI Slop) of variable quality. In every organisation I have met, the net improvement in organisational productivity or cashed efficiency savings is zero. Being generous, we have raised the bar of competitive parity.

The wider data agrees. Mckinsey 2026 survey found 80% of respondents said AI improved their individual productivity, but only 37% reported AI contributing to their EBIT, a share essentially unchanged from the prior year. Hours saved and spent on nothing are not productivity. They're slack.

🎯 The 10/10/80 reality
AI Tech is 10–20% of the solution. Good data another 10–20%. Reinventing how the work actually gets done is 70–80%. Most organisations fund the 1st, nod at the 2nd and ignore the 3rd.

🎯 Every model, dashboard, deck and email summary exists to drive a Decision, which leads to an Action, which delivers a Result.

Once, a few people made the real decisions. Now many people feed them information(overload). Its too easy to ask for another deck, another proposal or another insight to control risk, manage cost or "ensure a better outcome."

All those people and all that technology drive fewer decisions, less action and weaker results. (Copilot just produces the decks faster).

🎯 Start from the end
Flip it. Start with the Result, and ask how fast you can make the Decisions that drive it. Then be honest about where you need humans in the loop, where humans and AI should work together, and where you can take humans out of the loop.

Take your people with you. Firms where employee sentiment toward AI is positive are the firms showing real productivity gains. Resistance to change kills the 80% faster than any technical issue.

This takes time, focus and action. Done well, it delivers step-change results, not twenty minutes here and there.

Where is your organisation spending its 10%, and who owns the 80%?