Fictional client
Nobody complained, because nobody was short by enough to notice — thirty-eight dollars on an average shift, produced by a rule that prices a whole night at the rate for the day it started. 4,040 shifts last year and $169,220 of wages, and the same rule has been in the pay engine since it was configured in March 2022, which makes it $612,000 before superannuation or interest. Healthcare staffing is a business where the award is the product, and the software either prices every hour correctly or it quietly builds a liability nobody can see.
Merrivale Health Staffing does not exist. We cannot show live client systems, so this study is built around a fictional agency — invented firm, invented consultants, invented nurses, invented providers and invented figures. What is not invented is the work: the way rates attach to hours, the way a misconfigured rule survives for years, the coverage obligations providers carry and the failure modes are all real, and the screens are genuine working pages rather than pictures of pages. Every mockup on this page opens and responds to clicks.
On scope: this is written for healthcare and disability staffing in Australia, with a Melbourne agency as the example. Every rate on every screen is invented for illustration and is not an award rate. Classifications, loadings, allowances and penalty structures differ by award, by state and by enterprise agreement — in a real build those are configuration rather than code, and they are maintained against the instrument, not typed in once. All figures are Australian dollars.
Merrivale places registered nurses, assistants in nursing, allied health and disability support workers across metropolitan Melbourne and Geelong. Fourteen hundred workers on the books, 41,200 shifts filled last year, $21.4m of revenue, eighteen consultants and five people in payroll and compliance. Its problem is not winning providers or finding workers. It is that the thing it sells — an hour of care, priced under an award — is calculated by a system nobody has audited since it was switched on.
Healthcare staffing has a failure mode most businesses do not: the mistake produces a document that looks correct, goes to the person it harmed, and is filed by them without comment.
Two ideas, and both of them are things this business already believes and cannot act on. First, an hour is priced for the day it falls on. Every shift is split at midnight before anything is costed, so a night that crosses a day boundary is two priced segments that print as one line. There is nowhere to apply the rule by hand, so there is nowhere to forget it.
Second, the same engine that pays the worker costs the shift. Margin by provider and by day type is a calculation, not a number somebody typed into an agreement in 2024 — so a flat-rate client goes red on its own when rates rise on 1 July, without waiting for anyone to review it.
Seven screens follow. Each one opens as a real page — click through them.
The shape of the day: 142 shifts across nineteen sites, six still open, and the pay run sitting behind them with 78 exceptions held. The exceptions are the interesting part — this run has not been finalised, so those 78 are still correctable in a way that the 3,140 already paid are not.
One shift, decomposed into the bands the award actually pays on, with what was paid beside what should have been. Two hours on Saturday at the Saturday rate, six and a half on Sunday at the Saturday rate, and $54.60 that never appeared anywhere.
Underneath it, the same rule shown year by year since the payroll was configured — because sixty-one dollars is not the problem, and the rule that produced it is.
Five rules in the engine that produce a result the firm cannot defend, each with the shifts affected, the average error, the twelve-month cost and the date it started. Then the same exposure by provider, which shows why aged care carries two-thirds of it.
The overpayment sitting at the bottom of the table does not offset the underpayments, and the screen says so.
What the agency is relying on, separated into what it checked against the body that issued it and what it simply believes because a worker uploaded a document. Five of the seven are verified against a register; two are certificates on file, and the screen distinguishes them.
A second tab shows hours by week — a nurse who has gone from 32 hours to 57 over eight weeks, entirely in night shifts, without a single booking crossing a limit.
Tonight across nineteen sites, sorted by obligation rather than by start time. Each residential site shows how much of its night has a registered nurse booked, so a gap reads as a gap rather than as a row on a list.
The open shifts underneath carry the reason each worker declined — including nine declines on rate for a single sleepover, which is a rate conversation rather than a recruitment one.
One flat-rate agreement broken into day types, showing 23.4% on weekdays, 7.1% on Saturdays and a loss on Sundays and public holidays. Sixty-one per cent of that client's shifts fall in the two columns where it loses money.
Every provider then gets the same test. Both flat-rate agreements sit at the bottom, which is not a coincidence and not a negotiating failure — it is what a flat rate does to a business whose costs are set by the calendar.
The same decomposition, on the worker's phone. Saturday night shown as two lines because it was two days, with the rate against each. Offers carry an estimated pay figure produced by the pay engine rather than typed by a consultant, so the number on the offer is the number on the payslip.
This is also the cheapest audit the agency has: a pay engine that shows its working is checked by fourteen hundred people every fortnight.
Four phases of two weeks each — eight weeks end to end. The engine that prices an hour goes in first, because the worker's pay, the provider's cost, the margin on an account and the exposure report are all the same calculation seen from different angles. Each phase ends with something the team actually uses, not a demo.
Weeks 1–2.
Weeks 3–4.
Weeks 5–6.
Weeks 7–8.
Eight weeks is possible because most of it already exists. The worker and entity records, document handling, obligation and expiry tracking, escalation rules, the messaging engine and the portal are running in production across six sectors already — including staffing, where the same worker-compliance spine is in use. What actually gets built for a healthcare staffing agency is the sector core: the award pricing engine, the coverage model and the margin calculation. That is the part this study is about, and it is the part we would spend the eight weeks getting right.
This took a fictional agency and made its process visible. The same exercise works better with a real one. Tell us how your business actually runs and we will build a working mockup of it — your awards, your providers, your brand — before anybody commits to a build.
Start a Conversation See other sector blueprints