Knitted Bag Demand Forecasting: How to Build Numbers You Can Actually Reorder Against
Knitted bag demand forecasting is not a market prediction; it is a reorder calculation wearing a forecast's clothes. The number a buyer needs is not how many bags the season will sell but how many must be committed before the 7 working days of sampling and the 30 days of bulk make a second order impossible. Take an illustrative case: a 4,800-piece knit tote programme across four colourways, a 58-calendar-day reorder cycle and a 6-week cover target. Those four figures — total, colourway split, cycle length and cover — decide the first order, and the market outlook only decides whether the programme runs at all.
The Forecast That Was Wrong in the Right Place
Take an illustrative case rather than a client story. A buyer commits 4,800 knitted totes across four colourways, sells 3,900 by week twelve, and calls the forecast a 19% miss. The interesting part is not the miss: 1,200 pieces were ordered in the wrong two colourways, and 900 of the shortfall sat in the two that sold out by week six.
That shape of error is the normal one for a knit programme, and it matters because the corrective action is not a better total. The total was within 19%; the split was off by roughly a quarter of the season's volume. A forecast built only as a single number will always look acceptable while producing exactly this outcome, because the aggregate hides the only error that a buyer can act on. The rest of this article builds the forecast in the shape the error actually takes: a total, a split across colourways, a cycle length and a cover target.
Why a Knit Forecast Behaves Differently From a Sewn One
A knit forecast is written against a programme that produces in weeks rather than in days, so the planning variable is the reorder cycle rather than the sales curve.
Three structural differences produce that. First, the shape is written to a stitch file before anything runs, so a specification change during the season is a new programming fee and a new sampling window rather than a pattern tweak. Second, the colourway is a yarn choice rather than a cutting choice, so a split correction late in the season means booking a dyelot, not re-cutting panels. Third, the cycle is short and fixed: 7 working days of sampling plus 30 days of bulk, then transit. A programme that treats those three as flexible will write a forecast that cannot be executed, and a forecast that cannot be executed is a marketing document.
The Four Inputs a Knit Forecast Actually Needs
A knit forecast needs four inputs and no more: a weekly baseline, a colourway split with an error band, a repeat rate, and the reorder cycle length — with the cycle length being the only one the buyer does not choose.
The baseline should come from the closest comparable style expressed per week, not per season, because the cover calculation is a weekly one. The colourway split is where error concentrates and should carry a band: in the illustrative case, a split of 30/30/20/20 with a band of plus or minus 18% on each would have flagged the risk before commitment. The repeat rate is the share of the season expected to reorder, and it sets how much stock can be safely committed up front. The cycle length is derived: 7 working days of sampling, 30 days of bulk, about 21 days of sea transit, so roughly 58 calendar days from approval to landing.
- Weekly baseline
- Sell-through of the closest comparable style per week, not per season, because cover is a weekly calculation
- Colourway split
- Share per colourway with an error band; a band of about plus or minus 18% is a prudent opening assumption
- Repeat rate
- Share of the season expected to reorder, which sets how much is safe to commit in the first drop
- Cycle length
- 7 working days of sampling plus 30 days of bulk plus about 21 days of sea transit, so about 58 calendar days
- Cover target
- Weeks of stock held at the reorder trigger; 6 weeks is a common starting point for a general retail bag
- Minimum per colourway
- MOQ 100 pcs per style and colourway, which sets the floor of any split correction
- Capacity ceiling
- 7 production lines at about 200,000 pcs per month, which sets what a booking can absorb
Colourway Splitting: Where Forecast Error Concentrates
Colourway is the single largest source of avoidable forecast error in a knit programme, and the reason is that a colourway is booked as yarn rather than cut as panels.
A sewn programme can reallocate between colourways late, because the fabric is dyed or printed and then cut. A knit programme books yarn against a dyelot, so a late reallocation is a new booking with its own minimum of about 300 kg. The planning consequence is that a knit split should be set wider at the top and corrected later, rather than being equal-weighted: if a buyer has no evidence for the split, ordering 40/25/20/15 and correcting after week four costs less than ordering 25/25/25/25 and discovering that one shade carried the season. The minimum is the second constraint — at 100 pcs per style and colourway, a four-colourway programme has a floor of 400 pieces whatever the forecast says.
MOQ 100 as a Forecasting Instrument, Not a Constraint
A 100-piece minimum per style and colourway is best read as a forecasting instrument: it lets a buyer buy information at low cost in the first drop and commit volume in the second.
The arithmetic is straightforward. Suppose the forecast for a season is 6,000 pieces across four colourways. A single 6,000-piece commitment puts the entire forecast error on one order and one date. A first drop of 1,600 — 400 per colourway, four times the minimum — puts 27% of the season at risk and produces four weeks of real sell-through data before the second drop is committed. The second drop then carries the cycle: about 58 calendar days from approval to landing, so it has to be signed roughly nine weeks before the stock is needed. That is the number the forecast has to be built around, and it is why a forecast without a date attached to each tranche is not a forecast.
Buy information first
A first drop at four times the minimum generates sell-through data that a single large commitment cannot produce at any price.
Attach a date to each tranche
Every drop carries about 58 calendar days from approval to landing, so a tranche without a date is a wish.
Widen the split at the top
Unequal weighting with a band attached costs less to correct than equal weighting, because a correction books a dyelot.
Respect the floor
Four colourways at 100 pcs each is a 400-piece floor, so thin splits below that level are not executable however good the forecast.
Reorder Arithmetic: Cover, Cycle and Trigger Point
The reorder trigger is the point at which remaining stock equals the cover target, and it is calculated backwards from the cycle length rather than forwards from the sales curve.
In the illustrative case, a 6-week cover target against weekly sell-through of about 325 pieces means the trigger fires when stock falls to roughly 1,950 pieces. At that moment the buyer has about six weeks of stock and about 58 calendar days, or roughly 8.3 weeks, of cycle ahead. The cover is therefore short by about 2.3 weeks, and the honest reading is that a 6-week target is not compatible with a 58-day cycle at that velocity: the cover target should have been about 9 weeks, or the first drop should have been larger. That mismatch is the single most useful output of a knit forecast, and it is invisible in a spreadsheet that only tracks units.
| Step | Value in the example | What it decides |
|---|---|---|
| Sampling window | 7 working days | When the specification is frozen for the season |
| Bulk production | 30 days | When the goods are ready to despatch |
| Sea transit | About 21 days | When the goods land at the destination |
| Total reorder cycle | About 58 calendar days | The lead time the cover target must exceed |
| Cover target | About 9 weeks at this velocity | Safety stock held at the reorder trigger |
| Trigger point | Stock equal to cover, about 2,900 pcs here | When the repeat order is signed |
Reading the Forecast Into Capacity
A forecast is only executable if it fits the capacity window it is booking against, and the conversion is simply quantity per month against the monthly output available.
Capacity here runs at 7 production lines and about 200,000 pcs per month. A 4,800-piece drop is about 2.4% of a month's output, which is a comfortable booking; a 60,000-piece drop is 30% of a month and needs to be booked against a named window rather than assumed. The second point is seasonal: capacity is allocated in the order it is booked, so a forecast that is right but late produces the same outcome as one that is wrong. For a buyer, the practical rule is to book the window when the forecast is signed rather than when the sample is approved, because the 7 working days of sampling sit inside the 30 days of bulk planning rather than before it.
A Worked Forecast With Named Variables
The worked forecast below uses named variables only, so a buyer can substitute their own figures and reach their own answer.
- Season forecast F = 6,000 pcs across four colourways, or 1,500 per colourway if weighted equally.
- First drop D1 = 1,600 pcs, or 400 per colourway, which is four times the 100 pcs minimum.
- Observed sell-through at week four S = 62% of D1, or about 992 pieces, implying a weekly rate of about 248.
- Revised season estimate = 248 times 20 weeks, which is about 4,960 pcs rather than 6,000.
- Remaining requirement = 4,960 minus 1,600, which is about 3,360 pcs.
- Cycle check = about 58 calendar days, and cover at 248 per week needs about 2,232 pcs of safety stock.
- Second drop D2 = remaining requirement plus cover minus stock on hand at the trigger.
- Split correction D2 = weighted toward the two fastest shades, subject to the 100 pcs floor per colourway.
- Booking = the window is reserved when the forecast is signed, not when the sample is approved.
- Result = a season committed in two tranches, with the second one sized by observed data rather than by the original estimate.
The point of writing it this way is that every line can be argued with. If the season estimate is wrong, the error is in one line and its source is named. If the cover target is wrong, the error is in one line and it is a policy choice rather than a guess. A forecast written as a single number has no such property, which is why it cannot be corrected — only replaced.
Correcting Mid-Season Without Breaking the Line
A mid-season correction on a knit programme has three routes — a split correction, a repeat on the existing file, and a new shape — and they differ in cost by roughly an order of magnitude.
A split correction books yarn against a dyelot of about 300 kg and adds roughly 10 working days; it leaves the stitch file untouched. A repeat on the existing file costs nothing in programming and runs the same 7 working days of sampling and 30 days of bulk. A new shape costs the USD 180 programming fee again and restarts the development calendar. Read in that order, the lowest-cost correction is always the one that changes the fewest things, which is why the forecast should be built to make a split correction sufficient: one shape, several colourways, and a split that can be re-weighted without touching the file.
What to Measure After the Season, Not Before
The output of a forecast is not accuracy; it is a better set of inputs next season, and there are four worth recording.
Record the weekly sell-through curve rather than the season total, because cover is a weekly calculation. Record the split error by colourway in percentage points, because that is the error that was actionable. Record the realised cycle length against the planned 58 days, because slippage there is the difference between a cover target that works and one that does not. And record the return rate, about 2.4% of units shipped as a planning figure, because a forecast that ignores returns is forecasting gross rather than net demand. Four numbers, recorded once, will do more for next season than any amount of market commentary.
Turn your forecast into a booking
Send the style, the colourway split you have in mind, the season length and the destination market. The knit programme team will return the reorder arithmetic — cycle length, cover target, trigger point and the capacity window to book against — with sampling at 7 working days and bulk at 30 days. Start on the contact page.
Frequently Asked Questions
What is knitted bag demand forecasting?
It is the calculation that sizes the first order and the reorder trigger for a knit bag season. It needs four inputs: a weekly baseline, a colourway split with an error band, a repeat rate and the reorder cycle length of about 58 calendar days.
Why does a knit forecast need a weekly baseline?
Because the cover target and the reorder trigger are weekly calculations. A season total cannot tell a buyer when stock will fall to the trigger point, which is the only date that actually forces a decision. Written in that shape, the forecast can be corrected line by line rather than replaced entirely when a season moves.
How long is the reorder cycle on a knitted bag?
About 58 calendar days: 7 working days of sampling, 30 days of bulk and roughly 21 days of sea transit. The cover target has to exceed that figure, or the trigger fires too late to prevent a stockout. Recording the curve rather than the total also gives next season a baseline that is already in the right unit.
What is a cover target and how is it set?
Cover is the weeks of stock held when the reorder trigger fires. It is derived from the cycle length rather than chosen: at about 248 pieces per week and a 58-day cycle, the target is about 9 weeks rather than the usual 6. Air freight shortens the transit leg to about 5 days and raises the freight line by roughly USD 1.24 per piece, which is the trade to price explicitly.
Where does forecast error concentrate in a knit programme?
In the colourway split. A colourway is booked as yarn against a dyelot rather than cut as panels, so a late correction means a new booking of about 300 kg rather than a re-cut. Setting it below the cycle is the most common forecasting error, because the trigger then fires after the stock is already gone.
What error band should be attached to a colourway split?
About plus or minus 18% per shade is a prudent opening assumption when there is no prior data. It should be replaced with the observed split error from the previous season as soon as that figure exists. Attach an error band of about plus or minus 18% per shade when there is no prior data, and replace it with the observed split error once a season has run.
How does the 100-piece minimum affect a forecast?
It sets a floor on any split. A four-colourway programme has a 400-piece floor regardless of the forecast, so thin splits below that level are not executable and should be consolidated. The band is also what makes a wide top-weighting safer than equal weighting, because a correction books yarn rather than re-cutting panels.
Should a season be committed in one drop or two?
Two, whenever the cycle allows. A first drop at four times the minimum produces four weeks of observed sell-through, so the second drop can be sized by data rather than by the original estimate. It also makes a small first drop viable, which is the lowest-cost way to generate the sell-through data the second drop depends on.
When should a capacity window be booked?
When the forecast is signed rather than when the sample is approved. Capacity is allocated in booking order, and a forecast that is right but late produces the same outcome as one that is wrong. The date of the second drop is fixed by the cycle rather than by the sales curve, so it should be diarised when the first one is signed.
How much capacity is available against a large programme?
Output runs at 7 production lines and about 200,000 pcs per month. A 4,800-piece drop is about 2.4% of a month, while a 60,000-piece drop needs a named window booked in advance. Booking early also leaves room for a mid-season split correction, which needs the same window as the original drop.
What is the lowest-cost mid-season correction?
A colourway split change, which leaves the stitch file untouched and books yarn against a dyelot of about 300 kg. A repeat on the existing file is next; a new shape costs the USD 180 programming fee again. Converting a forecast into a percentage of monthly output is the quickest way to see whether it needs a conversation or merely a booking.
How much does a custom shade add to a correction?
A custom shade books a dyelot minimum of about 300 kg and adds roughly 10 working days. A stock shade carries no commitment, which is why stock shades are the safer choice for a split that may be corrected. Ranked by cost, the three routes differ by close to an order of magnitude, which is why the range should be built to make a split change sufficient.
Should returns be included in the forecast?
Yes. A return allowance of about 2.4% of units shipped converts a gross forecast into a net one. A forecast that ignores returns consistently over-commits the second drop by the return volume. Where a custom shade is part of the design language, book the dyelot once against the full season rather than per drop.
What should be recorded after the season ends?
Four things: the weekly sell-through curve, the split error by colourway in percentage points, the realised cycle against the planned 58 days, and the realised return rate against the 2.4% planning figure. Recording the realised rate after the season gives next year a figure that is specific to the brand rather than to the category.
People Also Ask
How do I forecast demand for a knitted bag?
Build it as a reorder calculation: a weekly baseline, a colourway split with an error band, a repeat rate and a cycle length of about 58 calendar days, then derive the cover target from the cycle.
Why did my knitted bag season sell out in two shades only?
Because colourway error concentrates in the split rather than the total. Equal weighting hides the risk; a wider top-weighting with a band of about plus or minus 18% costs less to correct.
When should I place a repeat order for knitted bags?
When remaining stock equals the cover target, which must exceed the cycle. At about 248 pieces per week, cover of roughly 9 weeks means the trigger fires at about 2,232 pieces on hand.
Can I change colourways mid-season?
Yes, and it is the lowest-cost correction available. It books yarn against a dyelot of about 300 kg and adds roughly 10 working days, while leaving the stitch file and the existing specification untouched.
How many pieces should the first drop be?
Small enough to be wrong at low cost and large enough to generate data. Four times the 100-pcs minimum per colourway is a workable rule, which on a four-colourway programme means about 1,600 pieces.
Sources and further reading
Every figure on this page is either our own programme data or attributed below. Open the originals before you write them into a specification.
- ISO 2859-1 — Sampling procedures for inspection by attributes, Part 1: AQL-indexed sampling schemesISO Sets the release inspection that sits between bulk production and despatch in the reorder cycle.
- ISO 13934-1 — Textiles, tensile properties of fabrics, Part 1: maximum force using the strip methodISO The specification threshold a repeat order has to hold against, which is why a repeat needs no re-testing programme.
- ISO 12947-2 — Textiles, determination of abrasion resistance, Part 2: determination of specimen breakdownISO The abrasion threshold referenced when a repeat is placed on an unchanged specification.
- Incoterms rulesInternational Chamber of Commerce Determines where transit risk transfers, which fixes the transit leg inside the reorder cycle.
- Standards and recycled fibre guidanceTextile Exchange Referenced for the fibre and certification choices that a colourway correction has to keep consistent.
- OEKO-TEX STANDARD 100OEKO-TEX The per-material certificate that stays valid across a repeat when material and market are unchanged.
Forecast the Cycle, Not the Market
A knitted bag forecast earns its keep when it is written as four named variables rather than one number. The total can be 19% wrong and the season still works; the split, the cycle and the cover cannot be wrong without consequence, because they are the three that decide whether the second drop arrives before the first one runs out.
For the cost side of the same decision, see bag programme cost planning. For the stock decisions that follow the trigger, see knitted bag inventory planning, and for the field the forecast is competing in, see knitted bag competitive landscape.