Traceability is becoming part of the print
# Traceability is becoming part of the print The code scanned on the printer's bench. It didn't scan in the retailer's warehouse. Both facts were checked with the same pack, a jar of sauce with a 2D mark on the label and a lot number beside it. The brand called it a systems failure and
Traceability is becoming part of the print
The code scanned on the printer's bench. It didn't scan in the retailer's warehouse. Both facts were checked with the same pack, a jar of sauce with a 2D mark on the label and a lot number beside it. The brand called it a systems failure and asked for a meeting with a software firm. The printer asked them to bring the jar. The mark was there. Nobody at the far end could read it. That's a printing failure, whatever else is also wrong with the database.
Brands want a pack they can identify later. A code, a mark, a batch, sometimes a unit. The printer doesn't own the database those things point at. The printer owns the mark on the substrate: where it sits, what it was printed with, whether a varnish or a laminate has buried it, whether the curve of a jar or the stretch of a film has distorted it, whether the file was this job's file. Traceability that arrives as a brochure about visibility still has to survive a die-cut, a glue line and a damp crate. If it doesn't, the identification is a story the pack can't tell.
The jar that passed in the office
He'd signed the flat label. That was the mistake, and he knew it by the time the warehouse photographs arrived. On the bench the code sat in a clear patch, contrast looked strong, their verifier beeped, and the quiet space around the mark hadn't been crashed by a descender or a die. He'd treated the label as the product. The product was a labelled jar, then a shrink, then a tray, then a week in a crate. The brand wanted to identify the pack later, not the sheet he'd found easiest to check.
They'd printed the code with the artwork, in the same pass, because the brand didn't want a second operation and the job was a steady one. A lot number was added at the filler, inkjet, into a second patch. Two marks, two moments, and the meeting had talked about them as if they were one "solution". They aren't. The printed code is only as right as the file and the ink. The filler's code is only as right as the head, the patch and the handling after it. A pack can carry a beautiful printed code and a smeared lot, or the reverse. Calling both traceability is how the wrong one gets blamed.
The warehouse failure had a chain, and none of the links was a server. The label went onto a curved jar, so the mark that had been flat became a mark on a radius. A varnish went over it because the brand wanted the panel scuff-resistant, and nobody had cut the varnish back from the patch. The jars were shrink-wrapped. The scanners at the retailer read at an angle, under their own lights, sometimes through the film, sometimes on bottles that were still cold and slightly wet. The verifier in his office had never been asked to do any of that. It had been asked to like a flat label. It did.
You could see the code. That was the detail that made everyone angry. A missing code is an obvious reject. A code that is present and unreadable gets into the supply and then fails in front of a customer who has done nothing wrong. The brand's first letter said the printer's data was bad. The data, when they finally looked, matched the job. The mark didn't survive the pack. He kept one of the failed jars, unscanned, on the shelf above the bench, because a photograph of an error message was persuading people it was a software row.
Where the mark sits is the job
A code is a printed object with neighbours. It needs a patch that stays flat enough, a colour behind it that isn't a photograph of herbs, and a margin that die-cutting, glue and a seam won't invade. On the sauce label the designer had put the code in a convenient hole in the layout, close to the shoulder curve and under the varnish plate. Convenient for the picture. Useless for a scanner on a round jar. He moved it, on the reprint, to a flat area lower down that had been occupied by a small icon nobody had defended in the meeting. The icon went. The code took the space. The varnish plate was cut back so that patch stayed open. None of that is a software setting. It's prepress, and it's a conversation the designer has to lose if identification matters more than the icon.
Ink is the next argument, and it gets waved through because the code "isn't really print". It is. A black that sinks into the face, or a black that sits on a laminate and flashes under a scanner's light, or a black that changes when the adhesive underneath is the wrong one, will all look like a code to the eye and then fail a read. He'd used the process black of the job for the first run because it was already on the press. Process black on that film, under that varnish, was not the black the verifier had liked once the varnish was there. The reprint used an ink they'd already run for codes on a plainer label, and the patch stayed unvarnished. He didn't promise every scanner in every warehouse. He promised a mark you could read on the packed jar in their own goods-out, with the film on, under a light that wasn't the bench lamp.
Stretch and seal wreck marks that were fine on the reel. A pouch that's printed flat and then formed can pull a code across the shoulder of a gusset. A shrink sleeve can move artwork after the code was verified. A carton crease through a 2D mark is a fold through the data. The press minder's job is to notice that the code in the file sits on the crease in the dieline, and to stop the job before a clever database is attached to a broken square. He'd missed it a month earlier, on a carton where the code looked fine on the sheet and split when the folder creased it. The brand could scan the flat. They couldn't scan the packed carton. Same failure as the jar, different fold.
The code that reads and lies
Readable and wrong is worse, and it doesn't look like a failure on a scanner. It looks like a success. The previous job's code, printed with the new picture, because the file name was nearly the same and the prepress operator dropped the old mark into the new layout. Every jar scans. Every scan points at the wrong batch, the wrong line, or the wrong product. The warehouse is happy. The recall, if it comes, is aimed at the wrong lot. He treats that as he treats a wrong language on a legal panel. It's a handover error on the press, not an IT incident. The fix is the same shape as any other file check. Someone reads the code's content against the ticket. A beep that says the square is a valid square isn't that reading.
The shift change is where that lie gets printed. The day file was still in the hot folder, the night file had a later time and an almost identical name, and the minder hung what was on the screen. The picture matched the proof on the bench from across the room. The code didn't, and nobody walked up to read it. He has since made the code a thing you read out loud at the start of the run, the same way you read a language line. If the person at the press can't say what the mark is supposed to contain, the press doesn't start. A database can't rescue a reel that left with the previous batch's identity baked into the ink.
Laminate glare is the quieter version of the unreadable mark, and it fooled them on a pouch before the sauce jar did. The code was sharp on the printed web. After the second film went down, the scanner at goods-out saw a shine and a partial read, then nothing, depending on the angle. You could still see the square if you tilted the pouch away from the lamp. The brand called that "intermittent", which is what people say when a fault has good manners. They stripped the laminate off the patch on the next file. The code sat in a window in the film, not under it. Identification that only works when you tilt the pack isn't identification.
A finished pack, or it isn't signed
He stopped signing flat labels for this brand. The OK is a packed unit. A jar with the label on, the shrink on, the code read in goods-out, the lot number checked on a filler sample if the filler will send one back. If they won't, he still won't sign a flat sheet as if it were the pack. He'll sign the label for the picture and write, on the same line, that the code hasn't been proven on the jar. Sales wanted that sentence left off because it looked like a doubt. It was a doubt. The doubt was the true part.
The buyer has to say where the mark will be read, or the printer is guessing a scanner he has never seen. On the jar, in a warehouse, through shrink, on a wet bottle, at a retailer whose lights aren't his bench lamp. Those aren't software requirements. They're the conditions of the print. If she can't answer them, the job can still be printed, and the ticket should say the check stopped at the plant door. Pretending the office verifier was her warehouse is how the first jar got signed.
He asks three things before prepress builds the patch, and he asks them of her, not of a brochure. What shape will the mark sit on once the pack is made? What film, varnish or shrink will cover it? Who reads it first, and with what, a person or a scanner? If the answer is a scanner on a curve, the patch moves to the flattest place the design can spare. If the answer is a person in a cold store, the type of the lot number has to survive being small, damp and hurried. If she wants both a printed identity and a filler's lot, they stay in two patches, so a smear at the filler doesn't take the printed code with it.
The reprint went out with the code lower down, varnish clear of it, the icon gone, and a note that the verifier result in the job bag was from a finished jar. He attached a photograph of the failed jar from the shelf, so the next prepress operator can see what a present-and-unreadable mark looks like. The software firm still had their meeting. He didn't go. Nothing in the job bag would have moved because a dashboard had a new screen. The file on the failed run had been the right file. The black had been the wrong black, in the wrong place, under a varnish, on a curve.
Brands will keep asking for packs that can be identified later. The printer's work is the patch, the ink, the position, and a check that happens after the pack is a pack. When the code is there and nobody can read it, the plan for finding that batch is a picture of a jar. He kept the jar. The flat sheet from the first run stays in the bag with a line through the signature, so nobody can pull it out and call it the standard.