Last week, a packer in the Adelaide Hills asked two of their most experienced QC staff to independently grade the same tray of baby spinach.
They disagreed on roughly one leaf in eight.
That isn't carelessness. Both graders had decades of experience. Both could explain exactly why a leaf was acceptable or rejected. They were applying a standard, just not the same one.
Because in fresh produce, the standard doesn't live in a system. It lives in people's heads.
Not on a screen. Not in a calibrated reference. In memory.
Where the cost actually lands
You won't find "disagreement" on a P&L.
You see it indirectly, under different headings, spread across the business.
A 10 to 12% over-rejection at intake shows up as lost yield. On a 10-tonne day, that's a tonne of saleable product written off before it even enters the line.
A similar under-rejection shows up later, as a credit note when a retailer pulls product early on shelf life.
And inconsistency between graders, the same borderline product called "in" one day and "out" the next, shows up as something harder to quantify: a customer who stops trusting your spec.
Most operators have an intuition for each of these costs in isolation. Few have ever treated them as one problem.
Why training doesn't close the gap
The first response to inconsistency is almost always training.
Recalibrate against a senior grader. Run alignment sessions. Tighten definitions.
It works. Briefly.
Then the senior grader goes on leave. A new starter joins. A different cultivar comes through. Fatigue sets in late in the shift, and borderline calls start to drift again.
Not because people aren't trained.
Because the standard has nowhere to live except in the heads of the people on shift at that moment.
A standard that lives in memory can't be audited, transferred, or held stable. It can only be approximated.
What changes when the standard moves
When the standard moves out of working memory and into a shared, calibrated reference, two things change.
First, disagreement converges toward that reference point. Not because it is inherently more "correct," but because it is consistent and external to any one person.
Second, the conversation becomes visible.
A rejection rate can be examined, not argued. A retailer challenge can be answered with evidence, not explanation. A new grader doesn't rely solely on shadowing. They can see how the standard is being applied in practice.
This doesn't remove the need for experienced graders. It changes their role.
Instead of holding the standard in their heads, they operate against a shared one, and make decisions about what to do with the outcome.
The follow-on
We've written previously that the industry is shifting from yield to margin.
This is the operational reality underneath that shift.
Margin doesn't compound on top of inconsistency. You can't contract or price your way out of a standard that moves by shift, by person, or by time of day.
At some point, the standard has to move out of heads and into something the whole business can point to.
More operators are starting to see that now, not as a technology decision, but as an operational one.