Field notes › Choosing software
Ask how the data gets out before you ask how it gets in
Every software demo spends time on getting your data in, because that is the part the buyer is nervous about and the part the vendor has practised. Almost nobody asks the opposite question, which is the one that determines how the next three years feel: how does the data get out?
It is not a hostile question. It is the same question as can I use my own information, and the answer shapes payroll, invoicing, client reporting and your ability to change your mind later.
Payroll is the export that runs every week
Whatever rostering system you choose, the hours worked need to reach payroll. If that path is a person retyping, you have bought a system that generates weekly manual work and weekly transcription errors.
Ask specifically: what comes out, in what format, and can it be produced without someone building a report each time. Then ask what happens when payroll changes its requirements slightly, which it will. A rigid export that exactly matches one payroll system is a liability the day you change accountants.
Invoicing needs the same data with a different shape
The hours that go to payroll and the hours that go on the client invoice are related and not identical: different rates, different rounding, sometimes different groupings. A system that can only produce one of them leaves the other as spreadsheet work, and spreadsheet work between two systems is where margin quietly disappears.
Worth asking whether per-person rates can override site or role rates, because in labour hire they routinely do, and a rate model that cannot express that produces invoices someone has to correct by hand.
Client reporting: push or pull?
Some clients want a report. More of them, lately, want access. Those are different products.
If the system can give a client site its own login scoped to their own roster, most of the reporting requirement evaporates, because they can answer their own questions whenever they like. If it cannot, you are committing to producing reports forever, and you should price that in.
The question everyone avoids: what if we leave?
Ask it directly, and watch how comfortable the answer is. Can you get your workers, their certificates and expiry dates, your client sites and their requirements, and your shift history out in a usable format, without paying for a project?
A vendor confident in their product will say yes without flinching, because they are not relying on hostage-taking for retention. A vendor who gets vague is telling you something useful about the next few years.
Our answer, for the record: it is your data, and you can have it. That is a low bar and it is surprising how often it is not cleared.
Integrations: be specific about the direction
Integration is a word that hides a lot. Pin it down by asking what moves, which way, and how often. Hours out to payroll weekly is a different thing from employee records in from an HR system nightly, which is different again from a live calendar feed a worker can subscribe to.
Be honest about which ones you actually need on day one. Most labour hire firms need hours going out and very little coming in, and buying a platform for integrations you will never build is a common and expensive mistake.
Watch out for exports that lose the point
A subtle failure: an export that gives you the roster but not the compliance context, or the certificates but not their verified state and history. What you usually need later is the pair, because the question being asked is who worked this shift and were they compliant at the time.
Ask to see a real export file, not a description of one. Five minutes with the actual columns tells you more than an hour of feature list.
The short list
Before you sign: see a payroll export file, confirm rates can be overridden per person, confirm whether clients get access or reports, ask what leaving looks like and get the answer in writing, and name the one integration you genuinely need in the first year rather than the five you might want.
The choosing rostering software post covers the wider evaluation, and implementation covers what the first month actually involves.