Read the Fair Work decision in McFadden v Mars Australia a few weeks back and it's worth a look if you're ever the one deciding who gets investigated, because the failure in it isn't really the investigation, it's the thing that happened before the investigation.

Short version. A team leader at the Asquith confectionery plant, 23 years there across Mars and Wrigley before it, no prior warnings, was dismissed over a food safety incident. A transcase holding around 485kg of hard-pressed peppermint mints was punctured, she and a food safety specialist inspected it and agreed roughly half the product should be binned, and then overnight it came apart. The operators binned about 80kg rather than the 240 odd that had been agreed. The electronic records in SAP and Nexus didn't line up with what had actually been done. Around 400kg was repacked and released for distribution before it was retrieved and destroyed. On the Commission's own description a number of steps in the procedure weren't followed at all, by a number of people. Mars investigated her.

She put in a detailed response, 17 pages of it, setting out where the operators, a leading hand and the food safety specialist had departed from the procedure as well. That got read as blame-shifting rather than as something worth going and checking. Deputy President Slevin found the inquiry was fixed on the team leader from the outset, found prejudgment in the fact that a mutual separation was floated before she had responded to the factual findings, and found there was no valid reason for the dismissal. She was reinstated with continuity of service and back pay.

The narrowing happened before the investigation started

Most of the commentary on this has framed it as an investigation that didn't follow the evidence where it went, and that's fair as far as it goes. But she was never in scope alongside anyone else and then narrowed down to. She was the scope. Someone decided, before an investigator picked it up, that this was a matter about her.

I've written terms of reference from the HR side and I've been handed them as an investigator, and the uncomfortable thing about scoping is that you do it at the point where you know least. Usually there's been a phone call, or a couple of paragraphs from an operations manager who has already formed a view about who's responsible, and out of that you have to produce a set of allegations against a named person before anyone has been interviewed. That's not carelessness, it's just the order things happen in. But it does mean the most consequential decision in the matter gets made on the thinnest information, and then it sets, because once it's in an agreed terms of reference, widening it means going back to the client and telling them the job they scoped isn't the job that needs doing. That's an awkward conversation and I suspect it doesn't get had as often as it should.

What I'd want out of a scoping conversation

Four things I'd push on now, having read this one:

• Who else touched this, and has anyone asked them anything yet, even informally.

• What does the procedure actually say should have happened at each step, and who owned each step. Not who's been named, who owned it.

• Is the person setting the scope the person the conduct sits with, or reporting to them.

• What's the trigger for coming back and asking to widen this, and who do I go to when I do.

The last one is the one that never gets agreed at the start and it's the one that would have mattered here. If there's no agreed route back, the default is to keep going with what you've got, and a competent process run inside the wrong boundary still produces a report that reads perfectly well.

When a respondent points at other people

The other part of this is harder to be confident about. When a respondent starts naming other people it often is deflection, and an investigator who chases every name they're given never finishes anything, so the instinct to discount it isn't wrong. The problem is you can't tell which kind you're looking at from the response itself. You can only tell by going and checking, and in this case the checking would have been quick, because the operators either binned what had been agreed or they didn't, and the system records were sitting there either way.

Worth noting the response ran to 17 pages, which is the kind of document that gets skimmed under time pressure and filed as the respondent being difficult. That's a judgement made in a busy week rather than a bad decision, but it's where this one turned.

McFadden v Mars Australia Pty Ltd [2026] FWC 2772, Deputy President Slevin.

Kyle Wilkins is the owner and lead investigator at Sage Investigations, an independent workplace investigations practice based in Sydney. This is general commentary on a published decision, not legal advice.