Capture once: one source for every transformation report

Most transformation teams produce the same facts several times a year, in different shapes. The EE report needs the workforce profile by occupational level, race, gender and disability. The B-BBEE management control calculation needs much of the same information. The skills development claim needs headcount, payroll and training records. The ESG report needs workforce and training figures again. The board pack needs a summary of all of it.

When each of these is built separately, from its own export, the figures drift apart. A person counted as senior management in one file appears as middle management in another. The headcount in the EE report differs from the headcount in the scorecard because the two were pulled on different dates. Nobody did anything wrong. The data was simply captured more than once. We see this in almost every verification season.

The fix is the principle we build every system around: capture once, use everywhere. It sounds like a software question, but most of it is agreement between people, and the software comes last.

Agree the definitions first

Before any data is organised, each term has to mean one thing. Who counts as an employee for each report? Which occupational level does each job grade map to? Which dates define the measurement period, and do they match your financial year? Is a supplier's B-BBEE level taken from the certificate valid on the invoice date, or from the one held at year-end?

Write the answers down in a short data dictionary that the transformation, HR, procurement and finance teams all agree to. It is unglamorous work, and it prevents more verification queries than almost anything else you can do.

Give every record one home and one owner

Each kind of data should have a single source and a named owner:

• People: employees, their occupational level and their EE declaration, owned by HR and drawn from payroll.
• Suppliers: each supplier's B-BBEE level, Black ownership, enterprise size and certificate expiry date, owned by procurement.
• Spend: procurement spend by supplier, drawn from finance and reconciled to the financial statements.
• Programmes: learners, bursars and enterprise and supplier development beneficiaries, with the support they received and their outcomes.
• Evidence: certificates, affidavits, agreements and proof of payment, each linked to the record it supports.

The owner is responsible for the record being right. Other teams read it. They do not keep their own copy.

Use identifiers that do not change

Names change and get misspelt. The same supplier appears in three forms across three systems. Every person, supplier and programme should carry a unique identifier that stays the same across payroll, procurement and the transformation record, so that matching is exact rather than guessed.

Where older systems have to be matched up, AI is useful: it can propose which records belong together and explain why. A person confirms each match once, and the confirmed link is reused from then on.

Keep history, not only the latest value

Transformation data changes during the year. An employee is promoted. A supplier's certificate is renewed at a different level. A beneficiary moves from enterprise development to supplier development. If your records hold only the current value, you cannot reproduce last quarter's report or show the verification agency what was true on a particular date.

Store each change with the date it took effect. Then close each measurement period with a locked snapshot, so the figures you claimed can always be reproduced exactly as they were.

Link evidence to the claim

A document in a folder proves little on its own. A document linked to the specific supplier, invoice, learner or contribution it supports, with its date and expiry, is evidence. When AI reads certificates and affidavits as they arrive, it can extract the details that count and attach each document to the right record, with a person confirming what it captured.

Produce every report from the same records

Once the source is organised, reports become views of it. The EE report, the scorecard, the skills claim, the ESG disclosures and the board pack all read from the same records, filtered and summarised differently. If a figure differs between two reports, it is because the definitions differ, and the difference can be explained in a sentence.

Privacy also becomes easier to manage. Reports that need only totals receive totals. Individual records are visible only to the roles that need them, and every access is logged. One controlled record is far easier to protect than a dozen spreadsheets travelling by email.

Where to begin

You do not need to rebuild everything at once. Start with the dataset that causes the most rework.

For many companies that is the supplier list, because procurement recognition depends on certificate details that expire at different times through the year. For others it is the workforce profile, because both the EE report and the management control calculation draw on it. Organise that one dataset well, connect the reports that use it, and then move on to the next.

We design and build transformation data hubs that bring payroll, procurement and training data into one controlled record, designed to be kept on the client's own systems with encrypted backups. If your team is producing the same figures more than once, write to us at sales@bsvision.co.za. We will start with the dataset that costs you the most time.