Line architecture: good, better, best and price tiers

Price tier as an axis, not a column

On most boards, the price tier is a label on the card — opening, core, elevated, or a number, sitting under the style name next to the category and the delivery. That is useful, and it is not architecture. A field on a card is storage, not architecture. It records what tier each style was assigned to; it says nothing about the shape those assignments add up to, because the shape only exists in aggregate and the card only ever shows you one style.

The architecture becomes visible the moment the tier stops being a field and becomes the grouping axis. Band the board horizontally by tier, column it by category or by drop, and the range resolves into a distribution, not a list. Distributions register by eye far faster than they read down a column: mass, proportion, and holes are apparent almost at once, where the equivalent judgment from a sorted table takes a pivot, a squint, and a willingness to trust the sort.

The distinction that matters to a merchandising lead is between two statements that sound similar and are not. “The range has three tiers” is true of nearly every range ever built — assign prices to enough styles and tiers emerge whether or not anyone intended them. “The range is architected” is a much stronger claim, and it has three parts: proportion (how much of the line sits in each band), distance (how far apart the bands are set, and whether the customer can perceive the step), and coverage (whether each category is present in the bands it should occupy). A range can pass on one and fail on the others. Most ranges that feel wrong in the room are failing on coverage while passing on proportion, which is exactly the failure a totals row cannot show you.

The practical consequence: the architecture read is a regrouping, not a rebuild. The same board holds both views — the category read the team works in daily, and the tier read that exposes the shape. Because it costs a regroup rather than a new artifact, it is worth doing at every review rather than once at range-plan sign-off. Tiers get assigned upstream, when the range is first laid out; that step is covered in how to build a line board. This guide picks up after the assignments exist and asks what they add up to.

What a healthy architecture looks like on the board

Banded by tier, a line that works has a recognizable visual signature. Each tier reads as a legible band with real mass — enough product that the band is obviously a part of the offer rather than a rounding error. One band should be visibly the widest. That band is where the brand lives, where the volume is planned, and where a customer describing the brand to someone else would land. Every category is represented in the bands it should occupy, and the color story deepens where the band is deepest rather than being spread evenly for its own sake.

The fastest diagnostic is a three-second test: someone walking up to the board cold should be able to say which tier the brand is built on. Not which tier has the best product, and not which tier the team is most proud of — which one the line is built on. If the shape does not answer that in a glance, the customer standing in front of the assortment will not answer it either, because they are doing exactly the same read with less context and less patience.

Which leads to the point that costs teams the most: architecture is proportional, not symmetric. An even split across three tiers looks orderly on a slide and almost always reads as a range with no point of view, because nothing is carrying the brand. Symmetry is what a range looks like when the tier structure was applied as a checklist rather than as a decision. The useful question is not whether the bands are balanced against each other but whether the widest band is the one you meant to be widest — a brand built on its core tier and a brand built on its opening tier are different businesses, and both are legitimate. Only one of them should be visible on your board.

There is a second read available in the same view, and it is the one that separates a decent architecture from a considered one: the ratio of option count to colorway depth inside each band tells you what that band is for. A band with many options in thin colorways is a browsing tier — breadth to give the customer something to move through, with the risk spread thin. A band with few options in deep colorways is a volume tier — conviction concentrated, and the depth committed behind it. Neither is right in the abstract. The question is whether each band’s shape matches the job you set for it, and a band whose shape contradicts its stated job is worth an argument in the room before the buy is built.

The gap: a tier band with nothing in it

The first failure mode is a hole. Visually it is unmistakable once the board is banded: a tier that is empty, or held open by a single orphan style doing the work of an entire band. The commercial consequence is straightforward — a customer who is ready to trade up out of the opening tier has nowhere to land. The trade-up either does not happen, or it happens at another brand, and in both cases the range has quietly capped what a good customer is allowed to spend.

The reason this is a visual job rather than a reporting job: gaps hide inside a category-sorted view. Read the line category by category — the way almost every team reads it, because that is how ownership is organized — and each category looks reasonably populated. Tops have styles at several prices. Bottoms have styles at several prices. Nothing looks wrong. The hole only appears in the cross-line read at a single tier, because a gap is a band, not a row. It is a property of the whole line at one price, and no single owner’s view contains it.

Two other flavours are worth naming, because they are more common than the plain empty band. The first is the category-level gap: the line as a whole spans three tiers, but a given category occupies exactly one — all the outerwear is elevated, all the tees are opening. The architecture is then inconsistent depending on where the customer enters. Someone who comes in through outerwear meets a premium brand; someone who comes in through tees meets a value one. Both are the same season, and the board shows the discrepancy instantly when you scan a band left to right and find whole categories missing from it.

The second is the phantom fill: a band that is populated, but only by carryover. The tier exists on the board, the counts look fine, and there is nothing new in it. For a customer who bought last season, that band is empty. This one is easy to miss unless carryover is visually distinguishable on the card, which is a good argument for making it so. What to do when you find any of the three is the same in kind: mark the band, then decide whether to extend an existing family up or down into it. Inventing a style whose only job is to plug a hole tends to produce exactly what it sounds like — a style with no reason to exist that still consumes development time, a minimum, and a slot in the buy.

The cluster: too much line at one price

The second failure mode is the opposite shape: one band swells until the others read as garnish. Visually it is a board with a dense middle and two thin edges, or a board where a single band runs three times the width of anything else. The line has a price it is really selling at, and everything else is decoration.

The mechanism is the interesting part, because nobody decides to cluster. Styles get priced one at a time as they are developed, each against its own cost, its own comparable, and its own margin target. Cost structures across a similar fabric base, a similar trim spec, and a shared vendor set converge, so the prices derived from them converge too. The cluster is an emergent property of pricing style by style — which is precisely why it is invisible to every person making each individual decision and obvious to anyone looking at the whole board at once. There is no meeting where the cluster was approved, and no report that flags it, because every input to it was individually correct.

The second-order version is subtler and more expensive: clustering within a tier. Twelve options at four adjacent price points is one band on the board and one price to the customer. Bands are ranges, not points, so distinctness inside a band is a real question — if two options share a price and a silhouette and a color position, they are competing with each other rather than with the market. On a spreadsheet those two rows are two entries with different style numbers. On a board they are adjacent tiles that look like the same product, and the redundancy announces itself before anyone has to argue for it.

The practical read is a two-pass scan, and the order matters. Scan within each band first, looking for repeated silhouettes at the same price — that is where the cheapest cuts live, because you are removing duplication rather than coverage. Then scan across bands for the proportion problem. Doing it in the other order tends to produce the wrong fix: the band looks overweight, so a style gets cut from the edge of it, when the actual redundancy was two tiles sitting next to each other in the middle. Sizing the total breadth that all of this has to fit inside is a separate exercise — the option count planner does that arithmetic, and this guide deliberately does not repeat it.

The inversion: price that does not track the product

The third failure mode is the strongest argument for doing any of this visually, because it is the failure only a visual board catches. An inversion is an option whose price does not track the product relative to its neighbours. Not wrong against cost — wrong against what the thing looks like next to what sits beside it.

It comes in two flavours. The first is the perceived-value inversion: an elevated-tier piece that reads cheaper than the core-tier piece next to it. Simpler fabric, plainer construction, less obvious make — and a higher price, because the cost happened to land there. In the spreadsheet, nothing is wrong. The cost is right, the markup is right, the row validates and rolls up cleanly into the category. The mismatch is between the price and the product’s appearance, which is information that exists nowhere in the data model. It exists only when the tiles are side by side, which is to say it exists only on a board.

The second is the structural inversion: a category’s opening price sits above another category’s core. The architecture then reads differently depending on which door the customer comes through, and the tier labels stop meaning anything consistent across the line. This one is a boundary problem rather than a neighbour problem — you find it by walking the line where two bands meet and asking whether the step between them is legible.

Inversions have two reliable sources, and both are worth checking by name. Carryover styles that kept an original tier assignment through a season where everything around them moved. And styles re-costed after a fabric substitution, a trim change, or a vendor switch, where the price was updated and the tier assignment never was. Both are bookkeeping-correct and architecturally wrong: the record is accurate, the shape is not, and no validation rule anywhere will object.

The read is two walks. Walk each band left to right and ask a single question of every tile: would this surprise me at this band’s price? Anything that gets a yes is either mispriced or misplaced, and the difference between those two answers is a real decision rather than a data fix. Then walk the boundary between each pair of adjacent bands and check that the step is legible — that the product at the bottom of the upper band visibly earns the distance from the product at the top of the lower one. If it does not, the tiers are set too close together and the architecture is doing less work than the price list suggests.

Tier balance against option count and colorway depth

Everything above reads the board by tile count, and tile count is the first reading, not the last. A tier’s real weight is its option count multiplied by colorway depth — and in footwear, multiplied again by the size run behind every style-colorway. Say an opening tier runs twenty options in two colorways while an elevated tier runs eight in six. The board looks lopsided by tiles, two and a half to one. The breadth-times-depth commitment is close to level. Which of those two numbers you are managing to is not a detail.

So read the board twice: once by tile count, to judge the shape of the offer as a customer meets it, and once by colorway swatches, to judge where the commitment actually sits. The second reading is the one that tells you what you have bet on. Teams that only ever do the first reading tend to be surprised at buy sign-off, when a band that looked modest on the board turns out to be carrying a disproportionate share of the units.

Colorway depth is also an architecture signal in its own right, not just a multiplier. Depth belongs where conviction is. A band carrying deep colorways is a band you are betting on; a band that is broad and shallow is a band you are testing. Both are defensible positions, and the board makes the position explicit rather than implied. The check worth running every season: if the deepest band and the widest band are not the same band, say out loud why. Sometimes there is a good reason — you are testing breadth at opening while betting depth on core. Often it is drift, and nobody had noticed the two had come apart. The colorway board view is the fastest way to see it.

Here is the boundary of what a board should be asked to do. What a given tier mix does to blended markup, planned margin, and markdown exposure is arithmetic, and it belongs on the plan, not on the board. Once the shape is set visually, take the tier structure to the numbers: run the markup implications through the IMU calculator and the exposure through the markdown calculator on retail-plan.com, and check how the tier mix reconciles against the numeric plan in line plan vs assortment plan. Set the shape here; prove the economics there. A board that tries to do both usually does neither well.

Footwear and accessories: what changes about the tiers

The architecture read transfers to footwear and accessories, but the unit of weight changes and so does the cost of getting a band wrong. How the board itself is structured for these categories — style by colorway by material, drop groupings, collection families — is covered in line boards for footwear and accessories. What follows is only what changes about the tiers.

In footwear, every style-colorway in a band carries a full size run, and frequently a width run alongside it. So a band’s weight is measured in pairs, not tiles, and an extra elevated-tier colorway is not one more card — it is a whole run. That makes tier decisions materially more expensive to get wrong here than in apparel, and it makes the second reading of the board (breadth times depth) not optional but the primary one.

Tooling is the second footwear-specific lever, and it has no apparel equivalent. Tiers often map to lasts and tooling, and tooling cost is amortized across the options that sit on it. An elevated band with one style on a dedicated last is a completely different commitment from an elevated band with four styles sharing one — the two look identical banded by tile count and are not remotely the same decision. Prebooks are the honesty check on all of it: the bands that wholesale actually prebooks tell you which parts of the architecture are real and which are aspiration you have been carrying since the range plan.

In accessories, there is no size dimension, so the architecture reads almost purely as style-color depth — which makes the tier bands unusually clean to read and the colorway judgment carry almost all the weight. Hero colors sit where the volume band is and go deep; fashion colors go broad and shallow across the bands that are testing. The distinctive accessories move is attach architecture: an opening-tier piece that exists specifically to attach to an elevated-tier hero — a cardholder against a bag, in the same leather article and hardware finish — is architecture doing a job rather than filling a band. It is only legible when the family is laid out together, which is an argument for holding both the family view and the tier view on the same board. Attach rate and UPT are the outcomes that kind of architecture is trying to buy, and neither is visible in a tier count.

Why architecture drifts, and what keeps it honest

Architecture is set early — in the range plan, before costing lands — and then eroded one style at a time. A fabric substitution moves a cost and a price with it. A vendor quote comes back higher than the target and a style crosses a band boundary. A style is cut at the review and the band it was holding open loses its only occupant. Each of those decisions is locally correct, defensible on its own terms, and made by someone with good reasons. Each of them also moves the shape a little, and nobody re-reads the whole distribution after any of them, because re-reading the distribution is a task nobody owns.

By the time the buy is built, the architecture on the board is frequently not the architecture anyone signed off. It was not overruled; it was accumulated away. The fix is structural rather than procedural — adding an architecture review to the calendar helps less than it sounds, because the drift happens between reviews and is invisible at the moment of each decision. Keep the tier as a live axis instead, so a cut or a re-cost re-renders the shape immediately and the room sees the consequence in the same meeting in which it made the change. That is a different failure mode from a board simply going stale, which is the broader problem covered in line board vs spreadsheet, Miro & Airtable.

The meeting mechanics — who is in the room, how the walk runs, how sign-off is recorded — are covered in how to run a line review; the addition here is simply to run the tier view as a second pass after the category walk, every time. And when the range is signed off, the architecture should carry forward into the assortment and the buy rather than being reconciled by hand from a deck. A connected board is what makes that possible, which is what Canvas, the visual line board inside RetailNorthstar, is built for — the board, the option counts, and the plan on a single shared data model, so a tier that moves on the board moves everywhere it is counted.

• ✓Line architecture is the price shape of the range — proportion, distance, and coverage across categories, not simply the presence of tiers.
• ✓Make the price tier an axis rather than a field on a card; the architecture is only legible as a distribution.
• ✓Three failures are visual: a gap is an empty band, a cluster is one band swallowing the line, an inversion is a price that does not track the product.
• ✓Tier weight is option count × colorway depth (× size run in footwear) — read the board twice, and the second reading is the one that counts.
• ✓The shape is set on the board and eroded style by style; keep the tier axis live so a cut or a re-cost re-renders the architecture in the room.

• How to build a line board, step by step →
• How to run a line review →
• Line boards for footwear and accessories lines →
• Option count planner — size a season’s breadth →
• Price tier (glossary) →
• Good, better, best (glossary) →
• IMU calculator on retail-plan.com →

Frequently asked questions

What is line architecture?

Line architecture is the price shape of a season’s range — how many options sit in each price tier, how far apart the tiers are set, and how that distribution runs across categories, colorways, and drops. Good, better, best is the most common shorthand for it. Every range has price tiers; a range is architected only when the proportion between the tiers, the distance between them, and their coverage across categories are deliberate rather than emergent.

What does good, better, best mean in apparel?

It is a three-tier price structure: an opening tier that gives the customer a way into the brand, a core tier that carries the volume and the brand’s point of view, and an elevated tier that sets the ceiling and signals what the brand is capable of. The labels are shorthand, not a rule — plenty of ranges run four or five tiers and some run two. What matters is that each tier has a job the rest of the line is not already doing.

How do you spot a price-tier gap on a line board?

Group the board by price tier rather than by category, so each tier reads as a horizontal band, then look for a band that is empty or held open by a single orphan style. Doing it visually matters because gaps hide inside a category-sorted view: read category by category and each one looks populated, and only the cross-line read at a single tier exposes the hole. Category-level gaps show up the same way — a category occupying one band while the rest of the line spans three.

What is a price inversion, and why does it only show up visually?

A price inversion is an option whose price does not track the product relative to its neighbours — an elevated-tier piece that reads cheaper than the core-tier piece beside it, or a category whose opening price sits above another category’s core. In a spreadsheet nothing is wrong: the cost is right, the markup is right, the row validates. The mismatch is between the price and what the product looks like, so it only becomes obvious when the tiles are side by side. Inversions cluster around carryover styles and styles re-costed after a fabric or trim change that kept their original tier assignment.

How does price-tier balance relate to option count and colorway depth?

Tile count is only the first reading of the board. A tier’s real weight is its option count multiplied by colorway depth — and in footwear, by the size run behind every style-colorway. Say an opening tier runs twenty options in two colorways while an elevated tier runs eight in six: the board looks lopsided and the breadth-times-depth commitment is close to level. Read the board once by tile count and once by colorway swatches; the second reading is the one that tells you where the commitment actually sits.

Should a line be evenly split across price tiers?

Usually not. An even split across three tiers tends to read as a range without a point of view, because nothing is carrying the brand. Most architected lines have one band that is visibly the widest — that is where the brand lives and where the volume is planned. The useful test on the board is whether someone walking up to it could say in a few seconds which tier the brand is built on. If the shape does not answer that, the customer will not answer it either.

How is line architecture different from pricing strategy?

Line architecture is a structural question — how the range is distributed across tiers, and whether that distribution holds across every category and drop. Pricing strategy is what each individual price should be, and the markup and margin arithmetic that follows from it. The board is the right surface for the first, the plan is the right surface for the second. Set the shape visually, then take the tier structure to the numbers to check what it does to blended markup and planned margin.