Five packhouse KPIs worth tracking (and three that waste your time)

Why a short list beats a dashboard

More metrics feel like more control, and in a packhouse they are usually the opposite. A dashboard of forty numbers takes longer to read than a shift lasts, so nobody reads it consistently, and the numbers that would have mattered sit next to a dozen that do not until the difference stops registering. The numbers below were chosen for one property: each one, on its own, points a supervisor at a specific action within the same shift. If a KPI cannot pass that test, it belongs in a monthly report, not on the wall.

The seven that follow split into two kinds. Packout, rework rate and cost per kilogram packed are yield and cost numbers: they tell you what the run cost and how much of it turned into saleable product. Throughput, downtime and on-time dispatch are flow numbers: they tell you whether the line and the dispatch schedule are running to plan. Labour cost per carton sits across both, and is usually the fastest way to notice that something is wrong before the other six confirm it.

The core seven, defined precisely

Packout percent: saleable output divided by input received, by weight or count, expressed as a percentage, tracked per run and per grower rather than blended. The trap is treating a single company-wide average as meaningful. Averaging a high-yielding crop with a low-yielding one produces a number nobody can act on, and a grower whose packout is quietly falling gets hidden inside everyone else's numbers until the problem is expensive.

Throughput per hour: units, cartons or kilograms, packed divided by direct labour hours on the line during the run. The trap is that it looks fine while the line sits idle for stretches of the shift and still hits target by running fast during the hours it does operate. Throughput needs to sit next to downtime before you trust it.

Labour cost per carton: total direct labour cost for the run divided by cartons produced, in currency per carton. The trap is that overtime pay makes this number spike on any day the run finishes late, even when throughput itself was fine. A spike here is a scheduling signal first, a productivity signal second, and reading it without throughput next to it will send you to blame the wrong thing.

Downtime: minutes the line was stopped divided by scheduled run minutes, as a percentage. The trap is that it only counts recorded stops. A line that never fully stops but trickles product all day under capacity reports a clean, low downtime figure while quietly missing its throughput target, which is exactly why the two have to be read together.

Rework rate: units re-graded or re-packed after already passing through the line once, divided by total units packed, as a percentage. The trap is labelling a rising rework rate a quality problem when it is usually a changeover problem, most often the wrong bag size or box spec loaded at the start of a run before anyone caught it.

Cost per kilogram packed: total run cost, labour, packaging materials and an overhead allocation, divided by kilograms packed, in currency per kilogram. Unlike the others, this is a genuine trailing indicator: it is the right number for margin conversations and quoting, and close to useless for a decision mid-shift, because by the time it is calculable the run is already finished.

On-time dispatch: loads that left within their scheduled dispatch window, divided by total loads scheduled that day, as a percentage. The trap is measuring it against carrier arrival instead of your own load plan. Measured the wrong way it credits the packhouse for a truck driver's punctuality and hides the packhouse's own scheduling slippage.

Three that usually waste time

Total volume packed, on its own: it flatters busy weeks and hides margin. Volume only means something read next to packout and cost per kilogram.

Average packout across all crops and growers: averaging a high-yielding crop with a low-yielding one produces a number nobody can act on. Keep comparisons within a crop and a grower.

Anything measured monthly and only monthly: a month-end spreadsheet tells you what went wrong three to four weeks too late to fix it. The useful cadence for the seven above is per run and per day; monthly is a rollup for trend-spotting, not the primary read.

Setting a baseline when you have no history

Most packhouses that start tracking these numbers have never formally tracked any of them, so there is no baseline to compare a first week against. The fix is not to wait for a clean quarter of data before acting. Pull four to eight weeks of receival, pack run and dispatch records you already keep, whether or not they were ever turned into a KPI, and compute the seven numbers retroactively. That retroactive pull gives you a rough starting range, not a target: early records were not captured with the discipline that live tracking imposes, so treat the first two to three weeks of actual live tracking, not the retroactive pull, as the real baseline.

Set the target as the median of that live baseline period, not the best day in it. A best-day target looks aspirational and gets ignored within a week because it was never representative of a normal run. A median target is one the floor can actually hit on an ordinary Tuesday, which is what makes a KPI worth checking rather than a number worth resenting.

Revisit the baseline after a full season, not before. Packout and rework rate both shift five to ten points between early-season and late-season fruit on the same line with the same crew, and a baseline set in week one of a season will read as a false decline by week twelve unless you know to expect the seasonal drift.

The cadence that changes behaviour, not just reports

Cadence matters as much as the numbers themselves. A five-minute daily read of yesterday's seven numbers against baseline, done at the start of a shift while there is still time to call a grower or adjust tomorrow morning's grading settings, changes what happens on the floor that day. A monthly PDF, however accurate, arrives after every decision it could have informed has already been made a different way.

Weekly rollups have a real place, but as a trend check, not the primary cadence: they are where you notice that a KPI has been drifting for three weeks running rather than having one bad day. The mistake is treating weekly or monthly as the reporting cadence and per-run tracking as optional extra data entry. Flip that, make per-run and per-day the primary read and the rollups a background check, and the same seven numbers stop being a compliance record and start being the reason a decision got made.

Reading three KPIs together: a worked example

Grower A's packout drops from 91 percent to 84 percent over one week. Read alone, that looks like a fruit quality problem on the grower's side, and the easy move is to call them about picking maturity. Read next to the other numbers from the same week, a different story appears.

Rework rate on Grower A's lots climbed from 3 percent to 11 percent over the same days. Downtime stayed flat at roughly 4 percent, so the line itself was not the bottleneck. Labour cost per carton rose 22 percent, driven by two hours of overtime most nights that week.

Put together, the pattern points at a scheduling problem, not a fruit problem: the line was running two SKUs on the same shift without enough changeover time between them, causing misgrading that required rework, the rework pushed the run into overtime, and the packout figure was picking up product that got sent out coded as waste during a rushed second pass, not fruit that genuinely failed grade. The fix is a changeover schedule, not a conversation with the grower about picking maturity, and packout alone would never have surfaced it. That is the case for reading these numbers as a set: any one of them, alone, supports a plausible explanation that turns out to be the wrong one.

Get them as by-products, not projects

Every number above falls out of records you already make: receivals, pack runs with graded outputs and waste, dispatches. If capturing them requires a separate reporting exercise, the exercise will be skipped in peak season, precisely when the numbers matter most. Fernable computes packout, throughput, labour cost per carton, downtime, rework rate, cost per kilogram and on-time dispatch live from the day's normal records, so the KPIs exist without anyone compiling them, and the daily read described above is a look at a screen, not a spreadsheet someone has to build first.