Your first job architecture is a bill, not a saving. Here is how to read it.

Most articles about job evaluation stop the moment the structure is finished. The roles are weighed, the bands are drawn, the PDF is ready. Congratulations. And that is exactly where the part nobody warns you about begins.

Because a job architecture is not a nice document. It is a measuring instrument. And the first measurement shows you something that was already true yesterday, just not on paper: people doing work of equal value for unequal pay. That list already existed. Now it has a date on it.

This article is about what happens next. Why a first job architecture almost always costs money rather than saving it, why you must not net the two sides of that difference against each other, and how to see exactly where you stand without talking yourself richer or poorer than you are.

Delivery is discovery

As long as you have no job architecture, "does everyone earn fairly?" is a feeling. Once you have one, it is a table. You put the band that follows from the weighting next to each role, and next to that what you actually pay today. At that moment your workforce splits into three groups: people inside their band, people below it, and people above it.

The group below is the reason the law exists. Someone does work that weighs, on the four factors (skills, effort, responsibility, working conditions), as equal to a better paid colleague, but earns less. That is precisely the difference the pay transparency directive wants to surface. You have now surfaced it yourself.

The group above is where it gets uncomfortable. And to understand why, you need to know how the repair works.

The repair only goes one way

Here is the heart of the whole story, and the reason a job architecture is a bill and not a saving.

Someone paid below their band has to come up. There is little argument about that: it is the entire point.

Someone paid above their band cannot simply be brought down. Pay is a term of employment, and you do not lower it unilaterally. That needs a written modification clause and a compelling interest, and in practice a phase-out arranged over several years. Without that, a cut rarely survives in court. This is not legal advice, but the direction is clear: up is fast, down is slow, hard, or not at all.

The result is a ratchet that only turns one way. Below-band pay gets lifted. Above-band pay gets frozen (the familiar "red circle": the salary stays put until the band around it has caught up) or slowly phased out. The sum of those two movements can only go one direction: up.

Why you must not net those two figures

It is tempting to think: we overpay some people and underpay others, so on balance it evens out. It does not, and it is an expensive mistake.

The people below their band cost you real money, in the short term. The people above their band gain you nothing in the short term, because you cannot lower them. One figure is a bill you are going to pay. The other is a risk you carry with you for years. They are separate. Do not add them together, or you credit yourself a saving that is not there.

There is another reason to keep them apart, and it is a human one. Research on pay transparency consistently shows that people below the middle of their group grow less satisfied and start looking for another job once they know their position. Above the middle, little changes. So the discontent lands precisely on the people you need to lift. Fixing it quickly is not only a matter of fairness, it is retention.

Why it says "at minimum", and not an exact figure

When you put your payroll next to your job architecture, you want one number: what does this cost me? Honestly, you cannot get that number out of your payroll, and anyone who says otherwise is talking themselves richer or poorer than they are.

The reason is privacy. A responsible analysis works with averages per role, not individual salaries. And an average does not tell you what the repair actually costs. A role whose average sits comfortably inside the band can still have people below the minimum that you need to lift. The average hides them.

What you can say honestly is a floor. Work out, per role, the least it costs to lift the average to the bottom of the band, add that up, and you have a figure the real bill always sits at or above, never below. Not a quote, but a floor. "At least this" is a defensible sentence. "Exactly this" is not.

See your own floor

Build your job architecture and import your payroll. The reading happens inside your own browser: only the per-role averages are sent, no individual salaries. You get back what you have to pay at minimum, how many people sit above their band, and the two figures kept apart, so you do not accidentally net one against the other.

Build your job architecture  or  see the example first

Is this bad news?

It feels like bad news, because you buy an instrument and the first thing it does is hand you a bill. But turn it around.

The bill existed before you started measuring. The gap between what your people earn and what work of equal value should pay was already on your payroll. You just could not see it. A job architecture makes an existing obligation visible, it does not create a new one.

And not measuring does not protect you. In the United Kingdom a large council ran into serious trouble, not because it had done job evaluation, but precisely because it had put it off for years while the gap kept compounding. That is a different context from a small or mid-sized business here, so do not carry it across one to one. As a lesson it holds: the liability grows whether you measure it now or not. The only choice you have is whether you bring it into view on your own timing and your own terms, or later under pressure when someone asks.

And "someone asks" is not theory. Any employee may ask what colleagues doing work of equal value earn on average, and you then have a deadline to answer. What that right involves is set out in what to answer when an employee asks what colleagues earn.

The bottom line

A first job architecture is a bill, not a saving, and that is exactly what it should be. It makes visible what was already there. The repair only goes one way: below-band you lift, above-band you cannot simply cut, so never net the two against each other. And because an honest analysis works with averages, the only honest figure is a floor, not an exact number.

If you want to know what each approach costs up front, from free to a consultant, read what a job architecture costs for an SMB. And if you want to see what such a structure looks like worked out, look at a real job architecture example. When you are ready to see your own floor, start with your first role. It is free, so you see the result before you spend anything.