Read the Room: Are we actually moving forward?

I can’t help thinking about how many lessons my career has taught me, about what makes change succeed and how to avoid repeating the same mistakes. I was part of implementing shared services in the 1990s. Yes, that puts me at the more experienced end of the scale! But I value every one of those experiences, and I hope they make me wiser and of more value than ever.

My real concern is that we still don’t seem able to get the message to the people who most need to hear it and understand the impact of each decision they make. You cannot design work without first understanding it, or without trusting the people who do it.

What I’ve been reading

Three pieces I’ve read recently make the same point from different angles.

Sergio Caredda, The Task Fallacy: AI business cases count tasks, but work is a system of hand-offs, checks, learning and relationships. One study he cites found AI coding agents raised coding activity by 240%, but software releases by only 30%. The task got faster; the work didn’t. His own shared-services story will be familiar to anyone who has lived through one: the work left behind in a lift-and-shift is the work nobody mapped.

Stefan Norrvall, Complexity Is Not in the Eye of the Beholder: “Leaving necessary work off the organisation chart won’t abolish it.” If no role, rule or system is designed to carry it, people carry it, until they can’t.

Phanish Puranam, Can Leaders Count on Self-Organisation? (INSEAD Knowledge): even the flattest, most “self-organising” organisations run on rules and norms somebody designed. The hand is hidden, not absent.

Start with the end-to-end

This matches everything I’ve seen in shared services, ERP implementations and automated transaction processing. You have to start with the end-to-end process:

• how the data needs to flow through it
• where it needs to land, and when
• how it serves all its masters
• and, from a governance perspective, that it isn’t collected too early or too late

You cannot short-circuit this by going straight to “core HR”, even with a shared service behind it. If the process doesn’t work, and the data doesn’t go where it’s needed when it’s needed, it will fall apart. You’ll get shadow systems, a loss of trust and very human resistance.

Finance systems and HR systems are different animals

Part of the problem, I think, is that many organisations are finance-led, and their leaders’ reference point is a finance system implementation. That’s a very different animal.

Financial regulation varies by region but shares a common backbone, so finance systems are like a kit house: you choose the paint and the carpets. HR systems come with the plumbing and wiring in, but you can move the internal walls, change what every room is for and redraw the floor plan.

The platforms are increasingly built for “adopt, don’t adapt”. The harder part is organisations being willing to simplify and standardise their own HR policies and processes first. Treat an HR implementation like a finance one and you’ll underestimate the policy, data and people work every time.

Not quick, but quicker than we make it

Is it difficult? Yes. Can it be done really quickly? In my opinion, no. But much of the delay comes from procrastinating over how to approach it. If we simply got on with it, using the best practice we already know, we’d reach the end goal sooner.

Three questions before your next change

Before the next restructure, system implementation or AI business case, ask:

• Where does this work actually live: in a role, a process, a system, or someone’s goodwill?
• What is each task quietly carrying, and where will that go now?
• Who has the authority to fix it when it stops working?

Design the work first. Then the structure.

What lesson do you keep seeing organisations relearn? I’d love to hear in the comments.

Credits and further reading

• Sergio Caredda, “The Task Fallacy”, The Intentional Organisation, Issue #61, 16 October 2026. sergiocaredda.eu
• Stefan Norrvall, “Complexity Is Not in the Eye of the Beholder”, Substack, 3 October 2026. Read here
• Phanish Puranam, “Can Leaders Count on Self-Organisation?”, INSEAD Knowledge, 16 July 2026. Read here
• Demirer, M., Musolff, L. & Yang, L. (2026), Writing code vs. shipping code: Productivity effects across generations of AI coding tools, NBER Working Paper 35275, as cited by Caredda.
• Juli Oliver-Smith, “Don’t Design the Job Yet. Design the Work.”, The People Process. thepeopleprocess.co.uk