Locking the line

What a lock actually commits

The reason sign-off matters is not that a group of people agreed. It is that a queue of work has been waiting for permission, and the lock is the permission. In most apparel and adjacent-vertical calendars that queue contains six things, and they start in roughly this order: sample orders go to the factory, style-colour records are created in the master-data system, the photography shot list closes and studio time is booked, the line sheet is laid out and distributed, sell-in dates are confirmed and the range goes to market, and the buy is placed.

This is why a change after lock does not cost “a bit of rework” — it costs whichever of those actions has already started. The distinction matters because it changes the shape of the cost curve. Teams reason about post-lock changes as though lateness is a smooth penalty: a week late is a little expensive, a month late is a lot expensive. It is not smooth. The cost sits flat, then jumps, then sits flat again, and the jumps happen on the days the downstream actions fire.

A style added the day before the shot list closes costs almost nothing, because it slips into a shoot that was already booked. The same style added the day after costs a returned half-day of studio time, an expedited sample so that something exists to photograph, and a re-cut of the shot list. Nothing about the range changed between those two days. Nothing about the style changed. One downstream action fired, and the price of the same decision changed with it — not in proportion to the time that passed, but in proportion to the cost of the thing that had already started. The worked example below puts arithmetic on that.

The operational consequence is that the release list has to be written down and visible, per lock, before the lock happens. When it is not, everyone in the room is estimating the cost of a change from memory, and the estimate is always the same one: “it should be fine.” The list is short, it changes little between seasons, and it belongs on the line review agenda next to the sign-off item rather than in someone’s head.

Most ranges need four locks and pretend they need one

A single sign-off has to be either early or late, and both are wrong. Lock everything early and colorways, costs and quantities are fixed before anyone has the information to fix them, so the season runs on a stream of exceptions that were really decisions taken out of order. Lock everything late and development spends months building against a range that can still be re-scoped, which is how sampling budgets get spent twice.

The fix is to lock different attributes at different times. Each lock fixes exactly one class of attribute and explicitly leaves the others open, which is what lets a team say “the range is locked” and be understood precisely rather than approximately.

Read the third column as the important one. Most arguments about whether a range is locked are really arguments about which attribute someone thought was fixed. A designer hears “locked” and understands the colorways are settled; a planner hears the same word and understands the option count is settled; a sourcing manager hears it and starts booking. Naming the lock removes the ambiguity at no cost.

Four is not a rule. A small range with one channel and no wholesale sell-in can run concept and range together, and a brand with a long lead-time component may need a fifth lock earlier than concept for that component alone. What does not vary is the principle: a lock names the attribute it fixes and the actions it releases. A lock that does neither is a status update.

The same add, on two dates

A merchant wants to add one style in three colorways to a range that has already passed range lock. The style is good, the gap it fills is real, and the request is reasonable. The question is not whether the style is worth adding — it is what the add costs on the date it is asked for.

Illustrative figures, chosen because they divide cleanly. Not benchmarks, and not drawn from any brand.

Take the add as 1 style in 3 colorways across an 8-size run, so 3 style-colours and 3 × 8 = 24 SKU records.

Case one: one week before range lock. No downstream action has fired. Samples are not ordered, master data is not created, the shot list is open, the line sheet has not been laid out. The work is to add three cards to the board, re-check the category’s option count and confirm the price tier still holds. One person, under an hour. The cost of the add is the cost of the decision — the meeting time to agree it — and nothing else.

Case two: three weeks after range lock. Four actions have fired. Now price the add by summing them:

• Samples. Two sample sets per colorway (development and salesman) = 3 × 2 = 6 sets, all placed off-cycle. At an illustrative 200 per set plus 100 expedite each, that is 6 × 300 = 1,800.
• Master data. 24 SKU records created outside the batch, each needing a costing line and a barcode. At an illustrative 15 minutes each, 24 × 15 = 360 minutes = 6 hours of data work, plus the error rate that off-cycle entry carries.
• Photography. The shot list closed and the studio day is booked. Three colorways added afterwards buy a returned half-day at an illustrative 1,200 — and the half-day is what makes the sample expedite non-optional, because the sample has to physically exist by the shoot date.
• Line sheet. Already laid out and distributed. A reissue is a layout revision plus a version notice, and the real cost is not the layout: it is that quotes are already going out on the previous version.

Direct illustrative cost: 1,800 + 1,200 = 3,000, plus 6 hours of master-data work and one line-sheet reissue with a live version problem attached. The buy has not been placed in this example, because buy lock is downstream of range lock — that is the one action that has not fired, and it is the reason this is still a manageable change rather than an inventory decision.

Same decision, two dates, opposite answers — and the interesting part is not the gap between one week before and three weeks after. It is that 1,800 of the 3,000 arrived on a single day. Had the request landed the day before the shot list closed, the 1,200 half-day and the 600 of expedite both disappear together — the expedite existed only to put a physical sample in front of the camera on a date that was already booked. The sample sets are still ordered off-cycle, so the add still prices at 6 × 200 = 1,200, but 1,200 is under half of 3,000, and the difference is one day and one action firing. The cost is set by which actions have fired, not by how late the request feels.

The four legitimate changes, and the gate each one passes

A locked range is not a frozen range. Markets move, a fabric fails, a key account asks for something, a factory misses a date. Change after lock is normal; uncontrolled change after lock is what destroys the board’s usefulness. In practice there are only four shapes of change, and every request is one of them.

Whichever shape it is, the request passes the same four-question gate before it is allowed, and the questions are asked in this order because each one can end the conversation:

• What comes out? A range has a planned option count and a plan behind it. An add with nothing coming out is not an add, it is a re-plan, and it should be treated as one. Requiring a name in this slot converts most casual adds back into the trade-off they always were.
• What has already fired? Read the release list for every lock the range has passed. This is the arithmetic from the worked example, and it takes minutes when the list exists.
• Who absorbs the cost? Not a rhetorical question. Sample expedites, a returned studio half-day and off-cycle data entry land in someone’s budget and someone’s week. Naming them before approval is what stops the same requester making the same request every season.
• Does the board still balance? Category mix, price-tier spread, colour story, delivery flow. A swap that holds the count can still leave the opening price point without an anchor, and a colour change lands across the whole board rather than in one card.

Four questions, and the fourth is the one that needs the board rather than a form. Checking whether a range still balances after a swap is a visual read across categories, tiers and colour — the kind of read a laid-out board makes in seconds and a change log cannot make at all. Use the line board checklist and the option count planner to make the first and fourth answers concrete rather than intuitive.

How the board and the buy quietly diverge

The failure worth naming explicitly does not look like a failure while it is happening. Changes accumulate informally: a colorway is dropped in a call with the factory, a delivery moves in an email with sourcing, a style is quietly swapped because the fabric did not pass testing. Each change is real, correct, and made by someone competent. None of them goes back to the board.

Six weeks later there are two versions of the truth. The board says what the range was agreed to be, the purchase orders say what was actually bought, and nobody can say which document is authoritative — because both are, for different questions, and no one wrote down which. The symptom is small and recognisable: someone asks how many options are in the range and gets three different numbers from three people, all of whom are reading a real source.

The damage arrives later, when the season is reviewed. Divergence does not cost much in the moment; it costs the ability to learn. If the board no longer matches what shipped, the two comparisons that make a season instructive both become impossible: the range at lock versus the range at receipt, and the range as planned versus the range as sold. The team ends up reviewing outcomes with no reliable statement of the intent that produced them.

The counter is not more governance. It is a single rule with a low cost: any change that would alter the range goes through the board first, even when the board is not where the change originated. Where the board and the downstream records sit on one data model this is automatic, because a dropped colorway drops everywhere at once. Where the board is a deck and the buy is a spreadsheet, it is a discipline, and disciplines survive only when the step is small — which is the practical argument for a connected board over a slide file that has to be maintained by hand.

Version the board, do not edit it

A board that is edited in place cannot answer the one question a lock exists to answer: what was committed, and when. Editing destroys the record silently — there is no error, no warning, just a board that now shows a range nobody ever signed. A month later the question “was this style in the range at lock?” has no answer, and the follow-up question “then who added it?” has no answer either.

So each lock produces a version, and a version has to carry three things to be useful: the date it was locked, the person who approved it, and the delta from the previous version. The delta is the part teams skip and the part that does the work. A list of styles is a snapshot; a list of what changed since the last lock, and why, is a record of decisions. The first tells you the state, the second tells you the reasoning, and only the second survives the people who were in the room.

The review value shows up at the end of the season. Put the range at lock next to the range at receipt and the differences are the season’s real story: the four styles that were dropped after lock and why, the delivery that moved twice, the colorway that was re-shaded and arrived a month late. That comparison is how a team finds out whether its lock dates are set in the right place — if half the range changes after range lock every season, the lock is too early, and the calendar is what needs fixing rather than the people missing it.

Practically, this means the locked version is a fixed object, not a state of the working file. Subsequent changes go on a new version that references the one before it. The working board can move as much as it likes; the locked versions do not move at all. Building that habit is easier when the board itself is designed for it — see how to build a line board for the underlying structure, and Canvas, the visual line board inside RetailNorthstar, for the same idea with the versions and the downstream records on one data model.

Exactly one person can unlock the range

One owner, named, usually the merchandising lead who owns the range, with the planner consulted on the balance question. The reason is arithmetic rather than politics: a locked range with two approvers is an unlocked range, because a requester who does not like the first answer will find the second approver, and the second approver will not have the release list in front of them.

The owner’s job is narrow and worth stating precisely. It is not to say yes or no. It is to answer with the cost — which actions have fired, what the change therefore costs, who absorbs it, and what comes out. Framed that way, most requests resolve themselves without a decision being needed, because a requester who wanted a style and did not know it meant a returned studio half-day and a line-sheet reissue usually stops wanting it, and the one who still wants it after seeing the number was right to.

Requests from outside the range team need one extra step. Sales, key accounts, marketing and leadership all have legitimate reasons to ask for a change, and none of them can see the release list. The escalation is not upward, it is inward: the request goes to the range owner, who prices it, and only then goes to whoever has the authority to spend that. Skipping the pricing step is how a range acquires changes that were approved by someone who was told the change was small — and whose informal approval is exactly how the board and the buy start to diverge.

Apparel: fabric is the lock behind the lock

In apparel the calendar lock and the real lock are usually different dates, and the real one is fabric. Greige booking, mill minimums and dye-lot commitments happen against expected volume well before the buy is placed, which means a style can be commercially open on paper while its fabric position is already committed. A team that reads only the published lock dates will assume a colour change is cheap right up until it discovers the dye lot was booked three weeks ago.

The practical adjustment is to add the material commitment to the range lock’s release list explicitly, so the board carries it. Then the question “can we change this?” is answered from the release list rather than the calendar, which is the only version of the question that has a reliable answer. Everything else on this page is written from the apparel case; the three sections below are where the mechanism changes shape.

Footwear: tooling makes the commercial lock final much earlier

Footwear breaks the ladder’s tidy sequence, because tooling is committed long before anything else. Lasts, outsole moulds and any style-specific tooling are cut per construction and per size run, they carry long lead times, and they are not recoverable if the style is dropped. That commitment lands in the development window — well before range lock in most calendars — which means that for a footwear range, the effective commercial lock happens at tooling, whatever the published calendar says.

The consequence is that the four changes are not equally available. Colour changes and material changes on an existing tool stay cheap late, because they do not touch the tooling. A new silhouette after tooling commitment has two options and both are constrained: build it on an existing last, which limits what the style can be, or order new tooling and accept a lead time that will not make the delivery. There is rarely a third answer, and pretending there is one is how footwear ranges acquire a style that arrives a season late.

So a footwear range needs its concept lock to do far more work than an apparel one, because concept lock is where the silhouette count is really being fixed. Ranges that share tooling across a family of styles get more late flexibility for free — a point worth designing for rather than discovering. There is more on how these boards differ in line boards for footwear and accessories.

Prebook: when the lock date is not yours to choose

Sporting goods and outdoor brands selling into dealers get the sharpest version of everything on this page, because the lock date is set by the market rather than chosen by the brand. The dealer prebook window opens on a date the trade calendar fixes, dealers write orders off the line sheet during it, and the range therefore has to be locked, sampled, priced and photographed before the window opens. Every lock upstream of that is back-planned from a date nobody in the building controls.

That changes the nature of a late change completely. In a purely internal calendar, a post-lock change costs money. In a prebook calendar it costs money and credibility, because the change lands on a commitment a dealer has already made: they wrote an order for a specific style, colour and delivery, and they have built their own floor plan and open-to-buy around it. A colour that shifts after prebook is not a rework item, it is a conversation with an account.

Two practical adjustments follow. First, the commercial lock has to sit meaningfully before the prebook window rather than up against it, because the line sheet has to be finished, not nearly finished. Second, the four-question gate gains a fifth question for anything after prebook opens: which accounts have already ordered this, and who is telling them. Model-year cadence and dealer-driven calendars are covered further in line boards for home, outdoor and equipment lines.

Accessories and bags: the material booking is the binding lock

For accessories and bags the binding commitment is usually material rather than tooling or fabric, and it binds earlier than teams expect. Leather is booked by hide lot because colour consistency across a lot cannot be recreated later. Hardware — zip pulls, plaques, clasps, branded furniture — is frequently custom-tooled and ordered against a minimum that covers several styles at once. Both are committed well before the buy, and neither is easy to unwind.

This inverts the usual ranking of the four changes. In apparel a colour change is often the cheapest late move; in bags it is frequently the most expensive, because a colour is a hide lot, and a hide lot is a commitment that cannot be split or resold as easily as a bolt of fabric. A drop can be cheaper than a re-shade. The reverse is true too: because hardware minimums are shared across styles, dropping one style can raise the per-unit hardware cost on the styles that remain — so the drop has to be priced across the family, not against the style leaving.

Before the commitment, and after it

Everything on this page is pre-commitment change control on the range: what the board is allowed to become between the first lock and the buy. Once the buy lock has fired, the range is inventory and the levers change entirely — reorder, transfer, hold, promote, mark down. Those are post-commitment inventory levers and they belong to a different discipline, covered in the in-season chase on retailnorthstar.ai.

The two share a construction — the same decision costs different amounts on different dates — and it is worth being clear about why. Here the cost is set by which downstream actions have fired; there it is set by how much of the season remains to sell into. Same shape, different mechanism, opposite sides of the buy. The handoff between the range decision and the depth decision is covered in line board vs assortment board.

• ✓A lock is a state change with a date, an owner, a version and an explicit list of downstream actions it releases — not a meeting outcome.
• ✓The cost of a post-lock change is a step function, not a smooth curve: it is the sum of the actions already fired, and it jumps the day each one fires.
• ✓Stage sign-off as four locks — concept, range, commercial, buy — each fixing one class of attribute and explicitly leaving the others open.
• ✓Four legitimate changes exist (drop, swap, colour change, delivery move) and each passes the same gate: what comes out, what has fired, who absorbs the cost, does the board still balance.
• ✓Changes made outside the board are how the board and the buy diverge — and divergence costs the ability to review the season at all.
• ✓Version the board at every lock rather than editing it in place; a version carries the date, the approver and the delta from the version before it.
• ✓Exactly one named owner can unlock the range, and the owner answers requests with the cost rather than with yes or no.

• How to run a line review, step by step →
• How to build a line board, step by step →
• Line review agenda template →
• Line board checklist →
• What is a line board? →
• Line board vs assortment board →
• The in-season chase (retailnorthstar.ai) →

Frequently asked questions

What does it mean to lock a line?

Locking a line is the state change at which the board stops being a proposal and becomes the input to a commitment. A lock is only real if a named list of downstream actions starts at it — sample orders, master-data creation, photography scheduling, line-sheet production, sell-in, or the buy itself. A lock has a date, a version, a named approver and a list of what it released. If nothing begins when the line is locked, nothing has been locked; a meeting simply ended.

What happens after a line review?

The signed-off range is versioned and the downstream actions the lock releases begin. Which actions depends on which lock the review closed: a range lock typically releases salesman samples, style-colour master data, the photography shot list and line-sheet layout, while price lists and the buy wait for later locks. The practical test of whether a review finished is whether someone can name what started that afternoon.

Can you add a style after the line is locked?

Yes, but the cost of the add is set by which downstream actions have already fired, not by how much the style itself costs. The same add requested a week before range lock costs the meeting time to decide it; three weeks after lock it can carry expedited samples, off-cycle master data, a returned photography half-day and a line-sheet reissue. Before allowing it, answer four questions: what comes out to make room, what has already fired, who absorbs the cost, and does the board still balance.

How many sign-off stages should a range have?

Most brands need four and pretend they need one: a concept lock that fixes styles and story while colorways stay open, a range lock that fixes options while costs stay open, a commercial lock that fixes cost and price, and a buy lock that fixes quantities. Collapsing them into a single sign-off means either locking everything too early, which forces changes that should have been decisions, or locking everything too late, which leaves development with nothing firm to work against.

Who can approve a change to a locked range?

Exactly one named owner, usually the merchandising lead who owns the range, with the planner consulted on whether the board still balances. One owner is the point: a locked range with two approvers is an unlocked range. When a change is requested from outside the range team — sales, a key account, marketing, leadership — it goes to that owner rather than to whoever is nearest the board, and the owner answers with the cost of the change rather than with yes or no.

Should the line board be edited after lock or versioned?

Versioned. Editing a locked board in place destroys the only record of what was committed, so a month later nobody can say whether a style was in the range at lock or added afterwards. Keep the locked version as a fixed object carrying its date, its approver and the delta from the previous version, and make subsequent changes on a new version. Being able to compare the range at lock with the range at receipt is what turns the season into something reviewable.