Employment equity reporting fails on data, not intent

Most employers who get employment equity reporting wrong are not resisting it. They are trying to assemble one return out of a payroll system, a spreadsheet of job grades, and a set of demographic records somebody captured at interview stage and nobody has touched since. The obligation is not the hard part. Producing a defensible number is.

That gap is about to be tested properly. Reporting opens on 1 September 2026 and closes on 15 January 2027, and manual submissions close on 1 October 2026. More importantly, this is the first cycle in which employers are actually measured against sectoral numerical targets rather than only filing a workforce profile.

The Employment Equity Amendment Act took effect on 1 January 2025. It removed the turnover threshold, so a designated employer is now any employer with 50 or more staff. It also set five-year targets across 18 sectors, running to 31 August 2030, at four occupational levels plus a 3% target for employees with disabilities.

The cycle that closed in January 2026 was a baseline. This one is the exam.

The commentary on all of this is already thorough. What it consistently skips is the operational question underneath: where does the data actually come from?

What a return actually asks for

An employment equity return is a reconciliation, not a form. To complete one honestly you need, for every employee, at a single point in time:

• Which occupational level they sit in
• Their race, gender and disability status
• Whether they are permanent, fixed-term or temporary
• What they are paid, for the income differential statement
• Every appointment, promotion, transfer and termination in the period

Read that list against how a fifty-person business is actually run.

Occupational levels usually live in a spreadsheet somebody built for a previous return. Demographics sit in a recruitment folder or in the HR system, but only for people hired after it was installed. Pay is in payroll. Movements are in nobody's system at all. They exist as email threads and an org chart that was last accurate in March.

Four sources, none of which agree, and a return that has to reconcile all of them.

Why the returns are late rather than wrong

The failure mode is specific and it is worth naming, because it is not what people assume.

Employers rarely submit false numbers. They submit late numbers, or they submit numbers assembled in a two-week scramble by one person who then leaves and takes the method with them. The following year it is rebuilt from scratch, slightly differently, which is how two consecutive returns end up telling stories that cannot both be true.

That inconsistency is the real exposure. A single return with a soft number is a data problem. Two returns that contradict each other is a credibility problem. It gets noticed precisely when you least want it to: during a verification, a tender submission, or a dispute.

There is now a harder consequence than embarrassment. The EE compliance certificate now depends on four things:

• Submitting the report on time
• Showing progress against targets, or a justifiable reason for missing them
• Complying with the National Minimum Wage Act for the preceding twelve months
• No CCMA finding of unfair discrimination in the preceding year Without that certificate you cannot contract with the state.

The consultation requirement makes it harder, not easier

The regime does not only ask for a return. It asks for evidence of consultation. The plan and the numbers must be discussed with employees or their representatives, not simply filed.

That obligation is reasonable and it is also the point at which a scrambled return falls apart. You cannot meaningfully consult on numbers you assembled last week and cannot explain. A consultation about a figure nobody in the room can trace is theatre, and everybody present knows it.

The businesses that handle this well are not the ones with better intentions. They are the ones who can answer "where did this number come from?" in one step rather than four.

What to do about it on Monday

The fix is unglamorous and it is not software procurement.

One. Write down where each required field currently lives. Not where it should live. Where it is today. Most businesses have never done this and are surprised by the answer.

Two. Pick the single system that already holds the most of it. That is usually payroll, because payroll is the one system nobody is allowed to let drift.

Three. Move the missing fields toward that system rather than buying a new one to hold them. Occupational level and disability status are two columns. The reason they are not there already is that nobody ever decided they should be, not that it is difficult.

Four. Record movements when they happen, not at return time. An appointment, promotion or termination captured on the day takes thirty seconds. Reconstructed eleven months later it takes an afternoon and it is a guess.

The test of whether this has worked is simple: on 1 September, can one person produce the return in a day, and can they explain every number in it without opening a spreadsheet? If yes, the consultation obligation stops being a risk and starts being a conversation.

The wider point

Employment equity is one of several returns a South African business owes. There are also B-BBEE, skills development levies, workplace skills plans, and for many, sector-specific reporting on top. They ask overlapping questions of the same underlying data, and almost every business answers each one separately, from scratch, in a different spreadsheet.

That duplication is not a compliance problem. It is an operations problem that shows up as a compliance problem once a year. That is why it never gets fixed. By the time it hurts, the deadline is too close to do anything but scramble.

This is the kind of work a discovery audit is for, and it usually lands in the workforce department. It establishes where the data actually is before deciding what, if anything, needs to change. In most cases the answer is not a new system. It is two columns, one decision about which system is authoritative, and a habit of recording things when they happen.