← systeric.com Interwise / Docs
Open App →

Job Architecture & Wage Structure

Job architecture is the system a company uses to decide what a job is worth before it decides what a person is paid. Without one, every salary is negotiated individually, and the result is a payroll that nobody can defend: two people doing the same work on different money, for reasons that made sense at the time and cannot be explained afterwards.

You already understand the idea from clothing. Nobody tailors every garment to every body. Sizes are defined, garments are made to them, and people are fitted to the nearest size with small adjustments. A company without job architecture is tailoring every single item, and discovering that it owns nothing it can compare.

The four layers#

Job families. Groups of related work: engineering, finance, sales, operations. A family is about the kind of work, not the seniority.

Levels. Seniority within a family, usually five to eight of them, from entry through to principal or director. A level is defined by the scope and independence of the work, not by the person’s tenure.

Grades. The pay classification a job is mapped to. Several jobs from different families can sit in the same grade if they are judged equivalent in value to the company.

Bands. The salary range attached to a grade: a minimum, a midpoint, and a maximum. The midpoint is the target for a fully competent person in that role, and it is usually anchored to market data.

negotiated one by one placed in bands
The same seven salaries, drawn twice. Nothing about anyone's pay changed. The only difference is whether there is a structure to say where each one sits and whether that position is defensible.

How jobs get graded#

Deciding which grade a job belongs in is called job evaluation, and the industry methods are all variants of the same idea: score the job against fixed factors, and let the total place it.

  • Korn Ferry Hay Guide Chart, the original point-factor method, scoring know-how, problem solving and accountability.
  • Mercer IPE, the most common in large Indonesian corporates.
  • WTW Global Grading System, which produces bands and grades directly.
  • Radford levelling, the default in technology companies.

They differ in the factors and the arithmetic. They agree on the principle, which is the part worth carrying: you evaluate the job, never the person in it. A brilliant person in a small job does not make the job bigger, they make a case for moving them to a bigger one.

The numbers people actually use#

  • Midpoint. The market rate for a fully competent holder of the job.
  • Compa-ratio. Salary divided by midpoint. Someone at 0.85 is paid below the target for their grade, someone at 1.15 above it. It is the single most useful number in compensation because it makes people in different grades comparable on one axis.
  • Range penetration. Where in the band someone sits, from minimum to maximum.
  • Market pricing. Anchoring midpoints to survey data. In Indonesia the usual sources are Mercer, Korn Ferry and the published salary guides.

The part that is law here#

This is where the Indonesian context stops being an implementation detail.

A wage structure and scale, struktur dan skala upah, is legally required. The obligation sits in Permenaker 1/2017, under the wage regulation PP 36/2021 as amended by PP 51/2023, itself under UU 13/2003 as amended by UU 6/2023. Employers must have one, must base it on job grading, and must communicate it to employees.

Meaning a business with forty staff in Bekasi has the same obligation as a bank, and no realistic prospect of running a point-factor evaluation to satisfy it. What they have instead is a spreadsheet somebody made once, or a consultant’s document from three years ago, or nothing at all and a quiet hope.

Why this matters to our strategy#

This is the same shape as the opportunity we already understand in payroll, appearing one layer up.

A legal obligation exists. The enterprise methodology built to satisfy it is far too heavy for our customer. So they satisfy it with a spreadsheet, badly, and re-check it by hand whenever anyone asks. That is precisely the pattern we exist to end, and it argues that wage structure is not an enterprise feature we would only need if we moved upmarket. It is an SMB compliance product hiding in plain sight, and one nobody has built for this segment.

It also completes the second half of our positioning. What will this hire actually cost me cannot be answered from a payslip alone. It needs a band: what the role should pay, what the total employer cost is at that level, and whether the offer being contemplated is defensible against the people already here. A company with no architecture cannot answer that question at all, which is why they answer it with whatever the last candidate accepted.

And it is a quiet retention tool for us. Payroll is switchable. A wage structure that every existing employee has been placed into, with history, is not, because migrating it means redoing the placement.

What this means for what we build#

  • Ship a structure they can adopt in an afternoon, not a framework they must configure. A small set of levels, sensible defaults for families that actually exist in Indonesian SMBs, and the ability to place people quickly. The enterprise version is a trap: getting it right takes a consultant, and our customer does not have one.
  • Grade the job, hold the person separately. This is the same positions-versus-people distinction that governs the org chart, and it must be right in the data model from the start. See Org Design.
  • Compa-ratio is the number to surface, because it makes unfairness visible at a glance across grades, and unfairness is the thing that gets a company into trouble.
  • Produce the statutory document. The output of this feature is not a screen. It is a wage structure and scale a company can show when asked to, generated from data they already keep.
  • Connect it to hiring. An offer being prepared should show where it lands in the band and what it costs in total, before it is sent, not after somebody has accepted it.