What your scale-ticket data is trying to tell you about yield

The dataset everyone ignores

A scrap yard generates hundreds of scale tickets a week. Most operators treat them as receipts - fire them off, file them, look at them again only when a supplier disputes a weight. But your scale-ticket stream is the highest-resolution dataset in your business: every load, every supplier, every material, every operator, every weight delta, every minute of the day. Everything you need to know about yield is already in there. You just have to read it.

Here is how.

Signal 1: Net-weight variance by supplier

Pull every ticket for a given supplier over the last 60 days. Plot net weight (gross minus tare) for the same nominal material code. You will see one of three shapes:

• Tight cluster. Their loads are consistent. They prepare material the same way every time. Pay them on time, treat them well, they are a flagship account.
• Wide cluster. Loads are wildly inconsistent. This is usually a peddler with a mixed source - sometimes industrial drops, sometimes garage cleanouts. Their average might be fine, but planning around them is hard.
• Slow drift downward. Net weight is trending down. They are either declining (selling somewhere else first, dropping the residuals at your gate) or their source is changing. Find out which.

The variance number is more useful than the average. Average says how big; variance says how predictable - and predictability is what fills your forward shipments.

Signal 2: Material yield per truck

This is the one that buys a new pickup truck.

Group tickets by inbound material code to outbound shipment. For every 1,000 lb of #1 Copper Bare Bright that comes in, how many lb of #1 Copper Bare Bright go out? If the answer is 970 lb you are running 97% yield, which is industry-good. If it is 920 lb, you are losing 80 lb of copper per 1,000 lb to something - moisture, dirt, mis-grading at intake, mis-grading at outbound, or worse, shrinkage.

When a yield number drops 3-4 points over a quarter, the cause is almost always at intake. Operators are calling the spec generously to keep suppliers happy and the scale moving. The fix is uncomfortable but cheap: re-train, photograph every borderline load, and rebuild the grade-call audit. The number comes back in two weeks.

Signal 3: Tare-weight drift on the same vehicle

Same vehicle, same driver, same supplier, 30 tickets - the tare weight should be flat to within a few hundred pounds. When you see a 6,000-lb tare in February and a 4,800-lb tare in June for the same Peterbilt 379, something has happened. Possibilities, in order of likelihood:

• They are weighing partially loaded (intentional or sloppy). Verify on cameras.
• The truck genuinely changed (new fuel tank, removed equipment).
• Your scale needs calibration. If the drift is across multiple vehicles, it is the scale.

The number tells you which. Aggregate the tare deltas across vehicles for a week - if they all move the same direction, calibrate.

Signal 4: Operator price-list discipline

Every ticket carries who priced it. Pull a week of tickets, group by operator, and look at price applied vs published price for the same material on the same day. The system should not allow a deviation, but real life leaks: manual overrides, expired discount codes, last-week prices that did not sync.

Operators with chronic overrides are not necessarily doing anything wrong - they may be reacting to a real condition (a competing yard up the road went up 5 cents, a strategic supplier needed a sweetener). But you should know who is overriding, by how much, and why. If you do not, you are running a price list that is more decoration than control.

Signal 5: Time-of-day curve

Plot ticket throughput by hour. Two patterns to watch:

• A flat middle. Healthy. Trucks are flowing.
• A 2pm crash. Either you are running out of scale-line capacity, or operators are slowing down post-lunch. Both are addressable. Capacity is a software config; energy is a staffing question.

A single yard that figured this out moved their second operator's lunch by 30 minutes and reclaimed an hour of effective scale-line capacity per day. That is one operator's salary in extra throughput, every year.

How to actually run this analysis

You do not need a data analyst. You need:

• Tickets exported to a flat table (CSV, Parquet, BigQuery - whatever).
• A weekly cadence - Monday morning, 30 minutes, same time every week.
• The five signals above on a single dashboard, refreshed automatically.

Scraplytics ships a Reports module that calculates all five of the above out of the box (Material flow, supplier scorecards, operator price discipline, time-of-day, vehicle tare). The hard part is not the math - it is the habit of looking.

The honest closing thought

I have spent six years inside ERPs that promised "AI insights" and watched operators ignore them. The yards that actually improved yield were the ones where the manager opened the same five charts every Monday and called the relevant person about anything that moved more than two standard deviations.

The dataset is already in your tickets. The discipline is the only missing ingredient.

Pick a Monday. Open the five signals. Call the right person. Do it again next Monday. Yield is built on Mondays.