And I mean that literally. Most of the grading systems still humming away inside organisations today, the fifteen-level ladders, the point-factor scoring, the tidy salary bands with their comfortable 20% spread, were built for a world where a “career” meant one employer, one office, one slow climb. You joined, you waited your turn, you moved up, and somewhere near the end there was a clock with your name on it.

That world is gone. The blueprint stayed. So we keep adding floors to a building designed for a different city, then act puzzled when the lifts don’t reach the top.

It assumed a lot of things nobody ever bothered to write down.

That everyone works from the same building. That careers only move up. That a job description might survive a whole year. That leadership sits neatly at the top of a pyramid instead of scattered across a matrix of dotted lines. And my personal favourite: that the person in the role is actually doing all of it, and not quietly handing half of it to an AI assistant that already wrote the first draft and ran the numbers before the coffee went cold.

Every one of those assumptions is now up for debate. Plenty of organisations still treat them as gospel. A framework can be mathematically perfect and organisationally fictional at the very same time, and most of us have signed off on one without noticing.

Picture two people. Both called “Manager.” Both on the same grade. Same band, same bonus target, same little box in the org chart.

Manager A runs a settled team of eight who’ve done more or less the same thing for years.

Manager B runs nobody permanently. She stitches together a different project team every few weeks. Half in the office, half remote, plus two contractors, a temp covering maternity leave, and now an AI agent she’s expected to “manage” like a junior who never sleeps, never complains, and occasionally lies to her with total confidence.

Same grade. Look me in the eye and tell me that’s work of equal value. Your evaluation model will, cheerfully, because it was trained to count headcount and budget, not to notice that one of those jobs barely existed five years ago.

Because here’s what the old systems are genuinely good at counting:

• Headcount
• Budget
• Reporting lines
• Formal authority
• Layers of hierarchy

And here’s what they mostly can’t see:

• Influence without authority
• Cross-functional orchestration
• Decisions made in the fog, without clean information
• Accountability that’s spread across a network, not owned by one desk
• Governing AI, and knowing when to overrule it
• Leading teams that assemble and dissolve in a matter of weeks

Reward vertical responsibility and only vertical responsibility, and you will underpay horizontal influence every single time. Which is a problem, because horizontal influence is how modern organisations actually run.

Then you’ve got the scale-ups and tech companies, where the job description isn’t really a document. It’s more of a rumour. Someone joins as a “growth marketer,” and by the end of the quarter they’re running partnerships, poking at the product, and quietly holding the data function together with tape and optimism.

Good luck grading that with a framework that assumes roles stand still. By the time the evaluation committee has finally agreed on the points, the job has changed shape twice and the person has picked up three skills nobody’s built a factor for yet.

We keep designing slow systems for fast companies, and then we look genuinely surprised when the roles don’t fit the boxes.

This is the one I can’t leave alone. Say you’re hiring remotely in India. So you buy “market salary data for India” and set your band. Reasonable. But which India, exactly?

A developer in Bangalore working for a US company earns dramatically more than a developer with the same skills, in the same city, working for a local firm. Same talent. Same postcode. Wildly different pay, decided entirely by whose name is on the invoice. So when you buy that data, whose salaries are you actually looking at? The local ones? The foreign-inflated ones? Some blended average that describes no living human being?

You’re not pricing a location. You’re pricing an accident of who happened to hire whom, and then stamping “objective and fair” across the top. Now run the same trick on your hybrid, remote and in-office people doing the identical job, and try saying the logic out loud without flinching.

Job architecture has always assumed the unit of work is a human role. That assumption expired a while ago, quietly, without a memo.

AI now does chunks of the research, the drafting, the analysis, the admin, the very things we once used to measure how “big” a job was. But that doesn’t shrink the human contribution. Usually the value just moves house. It shifts from producing everything to deciding what’s worth producing. From finishing every task to catching the ones that can’t be trusted. From processing information to actually exercising judgment. From effort to accountability.

So the most valuable person on the team may no longer be the one grinding through the biggest pile of work. It might be the one who knows exactly what must never be handed to the machine. Now tell me: where does that person show up in your evaluation factors? Because in most of them, they don’t.

Everybody reads the EU Pay Transparency Directive as a story about men and women. It’s much bigger than that. Strip away the legal language and it asks one gloriously simple, deeply uncomfortable question: can you explain, in plain words, why two people doing work of equal value are paid what they’re paid? Not just men and women. Anyone.

That isn’t a compliance question. It’s a fairness question. And fairness casts a much wider net.

You can hire the biggest names in consulting. You’ll get an elegant framework, a spotless audit, a watertight policy and a truly magnificent deck. And then some ordinary Tuesday shows up, someone looks at their band and their manager’s explanation of it, and asks the only question that has ever really mattered: “do I actually believe this?”

If the answer’s no, you didn’t fix anything. You laminated it.

Pay transparency was never really a reporting exercise. It’s a stress test for the logic underneath your pay decisions. The policy might pass. The spreadsheet might pass. The manager, standing there, mid-conversation, trying to explain the number, might not.

That’s the gap we work in at Growth People Pro, and it’s exactly why we refuse to treat this as a legal form to fill in. Compliance is the easy 40%. The hard part, the human and honestly far more interesting part, is whether your grades survive a real conversation with a real person on a real Tuesday. Whether your managers can explain a number without reaching for a script or suddenly discovering something urgent in their inbox. Whether people trust the logic, not just the paperwork.

So we take job architecture, reward and pay equity and treat them as one connected system, built around the humans who actually have to live inside it, for a world that keeps changing its mind every quarter. Employees experience all of it as one thing anyway, so that’s how it has to be designed. Less tidy than a fifteen-level ladder, yes. Also the only version that survives contact with reality.

Because job architecture was never really about the job. It’s about what a company is honest enough to say out loud about how it values its people. The legislation didn’t invent that question. It just took away our ability to dodge it.

So I’ll leave you with one honest question: if you published your salary bands tomorrow morning, would your own people believe them?

And if not, what breaks first? The grade, the band, or the person trying to explain it?