Home / Journal / Production tracking
Crochet production progress: four fields, three variances
Production tracking means turning the sentence "we are getting through it" into four fields a buyer can check — which operation, how many pieces sitting at it, who is blocking it, and when the next quantity ships. It measures how much of the order is still ahead, not how busy the floor looks.
Crochet has a long operation chain with a badly uneven time distribution. Preparation and crocheting move a lot of pieces quickly. Filling, assembly, shaping, metal detection and bagging move fewer pieces but take longer per unit. That shape produces a predictable distortion: a percentage counted in pieces reads optimistically, because it extrapolates from the fast end of the line. "Eighty per cent done" carries almost no information in a crochet order.
What follows is the shape of a weekly report that can actually be used to decide something. No fixed day counts — every order has a different operation mix, and a hard number in a template misleads more than it helps. Fields, tests and escalation lines only.
1. Why a percentage misleads · 2. The four fields · 3. Blockers come in five types · 4. Three variances and their triggers · 5. Photo evidence, four conditions · 6. Find the critical path first · 7. Three clauses worth writing · 8. Who reads which line
1. Why a percentage misleads
Take 1,000 plush pieces. The chain runs roughly: yarn preparation, crocheting, filling, assembly of trims, shaping, metal detection, bagging. When crocheting finishes all 1,000, a piece count says "1,000 of 1,000 done" and the order looks well past halfway. By labour hours it may not be. Filling and assembly are the hand-heavy stages: slow, hard to get right first time, and impossible to hurry by adding pressure.
| How progress is reported | What you see | What it means | Risk |
|---|---|---|---|
| Piece count | 1,000 crocheted of 1,000 | The front-loaded operations are finished | Reads optimistic; the tail blocks with no room left |
| Operation weighting | 4 of 7 operations, 52% weighted | Close to the real remaining workload | Needs the factory to disclose its operation and hour structure |
| Output dates | First carton: a date. Last carton: a date | Maps directly onto the buyer's receiving plan | Worthless if the date slides every week without comment |
The third is the most useful in practice: ask when the first carton ships and when the last one ships. The moment a factory commits to those two dates it has to schedule backwards, and scheduling backwards is what surfaces blockers. A percentage can stay vague; a date cannot.
2. The four fields
Four fields fit on one page. Few fields mean the factory will actually fill them in every week, the buyer will actually read them, and — usefully — the fields cross-check each other, so a padded number shows up at the intersection.
| Field | What goes in it | How to test it |
|---|---|---|
| 1. Operation | Which operation the order sits at, and which are complete | Line it up against the agreed operation list; no skipping |
| 2. Pieces in process | Pieces sitting at each operation, not cumulative output | The pieces in process should sum to roughly the order quantity; a gap means loss, rework or omission |
| 3. Blocker | The one thing holding it up, and who is waiting on whom | A blocker must name a party; if it cannot, it usually is not one |
| 4. Next output date | Date and quantity of the next shippable batch | Compare with last week's date; look for drift, not the absolute value |
Field 2 is the one most easily fudged and most easily tested. If the order is 1,000 and the report reads 300 crocheting, 400 filling, 200 assembly, 100 packing, it sums to 1,000 and is internally consistent. If it sums to 1,150, the extra 150 are either rework coming back through the line or last order's balance folded in. A number that does not add up is not a failure of the report — it is the question the report was supposed to raise.
Field 4 is about drift, not absolutes. One week's slip may be ordinary variation on a single operation. Three consecutive weeks each slipping a few days is a reasonably firm signal that the line has been filled with other work. The evidence is never in this week's number; it is in the series. Which is the whole argument for a weekly report rather than a one-off enquiry.
3. Blockers come in five types
Sorting blockers into types converts "it is held up" into something with an owner. In crochet orders, more than nine in ten fall into five categories, and each is fixed differently.
| Type | Typical wording | What to ask next |
|---|---|---|
| Material | Yarn has not arrived; shade not confirmed | Is the purchase order placed, what date did the mill give, is there a workable substitute shade |
| Subcontract | Embroidery shop has not returned it; printing queue | Date on the subcontractor's receiving note, queue position at their plant, can it come back in part batches |
| Equipment | Metal detector down; not enough pressing stations | Expected repair or changeover time; is there a spare or a loan unit |
| Labour | Short-handed this week; everyone is on another order | Which operation is short, how many people, when they arrive |
| Rework | A batch has to be redone | Cause, quantity affected, effect on dates, who carries it |
Subcontract is the hardest to control, because the factory is also waiting on someone. Embroidery, printing, heat transfer and custom webbing frequently sit outside the plant. The way to test this category is not to listen to the explanation but to ask for two documents: the date on the subcontractor's receiving note and the quantity returned. With those two, the outside step stops being a black box. Without them, the report is only repeating what a third party said.
Rework deserves a standing question rather than an occasional one, because it corrupts two dimensions at once. Reworked pieces leave the in-process count and come back later, so progress appears to stall, and factories rarely volunteer it. Put one fixed line in the weekly template: any rework this week, how many pieces, what caused it. Even a weekly "none" is worth the space.
4. Three variances and their triggers
Variance comes in three kinds. Mixed together they turn into an argument about trust; separated, each has a trigger and an action.
Quantity variance
The theoretical piece count implied by material issued does not match what is actually in process. Usual causes: loss above expectation, rejects not pulled out, or trial pieces mixed into bulk. A workable trigger is a written explanation once a single operation deviates by more than 3% — that 3% is our own internal threshold, not an industry standard, and both sides can set it differently before it goes into a contract. Catch quantity variance early, because replacing material usually takes longer than making the goods.
Time variance
The committed output date keeps moving. The test is not "a day late" but two consecutive reports with the date pushed back. Once is variation; twice is a trend. The question at that point is not "can you catch up" but "what would have to give in order to catch up" — normally splitting the shipment, or resequencing an operation. That question is what reveals where the real room is.
Quality variance
Rework or inspection failure rate crosses the agreed threshold. What makes quality different is its lag: this week's defects usually trace to a decision made two weeks earlier — a changed filling, a changed yarn, a new pair of hands. So the response is not to push harder, it is to stop and find the cause, because every piece made in the meantime is a candidate for rework. For inspection mechanics, see spec-to-bulk tolerance.
5. Photo evidence, four conditions
Photos are the cheapest evidence and the easiest to fake. One good finished-goods shot can serve ten different orders' weekly reports and the buyer will never know. Four conditions have to hold at once before a photo counts as evidence.
1. A batch card in frame. A card showing batch number, date and operation name. Its job is not decoration; it anchors the image to a specific batch, day and operation.
2. Wide and close, as a pair. The wide shot answers how many; the close shot answers how well. Wide alone hides the craft, close alone hides the scale.
3. A countable quantity. The strongest version is weighing: unit weight times piece count should approximate the pallet gross weight within a sane margin. This beats every camera angle because it does not depend on the photographer's honesty.
4. A fixed position. Same angle, same background, every week. A single photo shows nothing about trend; a fixed-camera series shows exactly how much more there is than last week.
A styled hero shot, by contrast, has close to zero value as progress evidence. It proves the factory can make a good sample, and that was settled at sampling.
6. Find the critical path first
The critical path is the longest serial chain of operations on the order. It sets the total duration, and it is where attention belongs. The common mistake is to watch the operation with the most pieces — usually crocheting — which can normally be run in parallel and staffed up. What actually holds an order is more often one of these:
The subcontract step. Embroidery and printing queue at the subcontractor's plant, not yours. Your floor can be fast and the order still cannot ship.
A specified material. A named yarn shade, a particular filling specification, custom webbing. These run longer than commodity materials and have no drop-in substitute.
Printed packaging. Cartons, stickers and hangtags are print work with a fixed span from artwork approval to bulk, and they are routinely remembered last.
The practical way to find the path: have the factory lay the operations out in sequence and ask one question — which single step, if it ran three days late, would make the whole order three days late. That answer is the critical path. Every subsequent progress question should centre on it, because compressing an operation that is not on the critical path does nothing for the ship date, which is the most common wasted effort in a rush.
7. Three clauses worth writing
Weekly chasing is not a system. Three clauses belong in the order terms. Note that all three govern information obligations, not lead-time promises — a fixed day count in a contract leaves neither side anywhere to go when a process goes wrong.
One: the reporting obligation. Name the four fields, the cadence (weekly during bulk), and the recipient. Specifying the fields does far more than "keep the buyer informed", which cannot be verified.
Two: the escalation obligation, with a response window. List what must be reported unprompted — critical-path slippage, rework, material shortage — and how quickly a recovery plan follows. This constrains how fast someone answers, not how fast the work finishes.
Three: who decides on variance. When dates slip, who chooses between splitting the shipment and moving the whole order; who pays the extra freight a split creates. Left unwritten, this becomes a dispute rather than a decision. For how changes are priced, see the OEM/ODM process.
8. Who reads which line
The same report is read three different ways. Designing it as one page is what lets each reader find their line in thirty seconds.
Purchasing reads the output date. It decides whether the launch moves and whether to book space earlier. No need for floor detail.
Merchandising reads the blocker. It is the only item that requires them to act. An empty blocker column means nothing to push this week.
Management reads the variance count. A single week says little; consecutive weeks say something real about how stable the line is — and that number is the best available input to whether the next order goes to the same place.
Our own version reports the four fields weekly, requires a batch card in every milestone photo, and marks the critical path at the scheduling stage. The piece on booking capacity ahead of Chinese New Year covers how to reserve the line; this one covers how to confirm the line is actually moving once reserved. Both are needed, and neither works alone.
Frequently asked
How often should progress be reported — is daily better?
Weekly during bulk, plus immediate notice at milestones, is usually enough. Daily reporting tends to distort rather than inform: crochet has long per-unit cycle times, so day-to-day output swings widely and normal variation reads as a problem. It also pushes the factory to optimise the report instead of the line. The only thing worth daily attention is the critical-path operation.
The factory only says "it is in production". How do I get further?
Stop asking how much is done. Ask four specifics: which operation now, how many pieces at that operation, which operation is slowest this week, and when the first carton ships. The first three are answerable; the fourth forces a backwards schedule. If none can be answered, scheduling has not been done — a worse signal than a slow line.
Why is a piece-count percentage inaccurate?
Because the operation times are unevenly distributed. Preparation and crocheting are quick per unit and move many pieces; filling, assembly, shaping, detection and bagging are slow per unit and move fewer. A percentage in pieces extrapolates from the fast end and therefore reads high. Operation weighting, or simply substituting output dates for a percentage, is more honest.
How do I track a subcontracted operation effectively?
Ask for two records: the date on the subcontractor's receiving note, and the quantity returned. Those turn an outside step from hearsay into fact. Also ask whether it can come back in part batches — a subcontract step that can be split carries far less risk, because one delay no longer takes the whole order with it.
Can a rush recover a slipped schedule?
A rush compresses queue time, not process time. Crocheting, filling and shaping are paced by hands and skill, not by urgency. It also displaces other orders on the same line, so the risk moves rather than disappears. The more useful question is whether part of the order can ship early enough to protect the launch date.
Can a live video inspection replace a weekly report?
No — they answer different questions. Video is a snapshot at one moment, good for confirming whether this batch is made correctly. A weekly report is a series, good for judging whether it will be late. Video cannot show drift; a report cannot show hand feel. Neither substitutes for the other.
The completion date in an inspection booking is only as good as the progress reporting behind it. Booking the inspection takes that date and works backwards to the cut-off, with the buffer written down before the order starts.
Related: labour minutes and standard time — how the time is measured and rated, what the allowance covers, and why the first hundred pieces are not the thousandth.
Want to see what our weekly report looks like? Send the operation mix and the destination market, and we will return a sample built on the four fields with this order's critical path marked.