Accordion Theory in Retail: The Cycles of Assortment Width Explained

Accordion theory holds that retail institutions move through cycles in which the width of the assortment expands and contracts, like the bellows of an accordion. General stores carrying a little of everything give way to specialists carrying a great deal of one thing, which in turn give way to generalists again.

Stanley Hollander set it out in 1966 as one of a family of cyclical theories of retail change. It is less famous than his earlier wheel of retailing, and it explains a different variable: the wheel is about price and margin, the accordion is about breadth.

What the theory claims

The claim is that assortment width is not a permanent choice but a position in a cycle, and that both ends of the cycle carry the seeds of the move back.

A wide, shallow assortment offers convenience. The shopper completes several errands in one visit and accepts a limited choice within each category. Its weakness is that it cannot satisfy a customer who cares about a particular category, and it cannot compete on depth with anyone who does.

A narrow, deep assortment offers authority. The retailer knows the category, carries the range, and can serve the demanding buyer. Its weakness is exposure: the business rises and falls with one category, and growth eventually requires either more stores or more categories.

The mechanism that drives the cycle is that success at either end generates pressure towards the other. A specialist that has saturated its category grows by adding adjacent ones, and becomes a generalist. A generalist under pressure from specialists in its most profitable categories responds by deepening those, and becomes a specialist in effect if not in name.

The historical sequence Hollander described

The illustration usually given runs through United States retailing, and the same sequence is visible in British high streets.

The nineteenth-century general store carried food, cloth, hardware and tools together because settlement density supported only one shop. As towns grew, specialists appeared: the butcher, the draper, the ironmonger, each able to survive on one category because the catchment was large enough.

The department store then reversed the movement, assembling many specialist departments under one roof and one name. Supermarkets and later hypermarkets widened further, adding non-food ranges to grocery until a single store carried clothing, electricals and pharmacy.

Category killers — the large specialist formats in toys, electricals, office supplies and do-it-yourself — reversed it again from the 1980s, taking depth in a single category to a level the hypermarket could not match. Many of those were then squeezed in turn, partly by online retailing and partly by the generalists returning to the categories they had ceded.

Why the cycle turns

Four forces recur in the explanations, and a good answer names the ones that fit the case instead of listing all of them.

Scale and catchment. Specialisation requires enough customers within reach who want that category. Growth in population, mobility and, latterly, online reach all widen the catchment and make narrower specialisation viable.

Operating cost. Wide assortments carry complexity: more suppliers, more stock-keeping units, more space, slower turns. When margins tighten, complexity is the first thing examined, and the response is usually to cut the tail of slow-moving lines, narrowing the range.

Competitive response. A generalist losing a category to a specialist can either concede it or fight on depth. Fighting on depth means becoming more specialist in that category, which narrows the effective assortment even if the store still stocks everything.

Customer time. The weight customers place on convenience against choice moves with how much time they have and how easy travel is. The convenience-store revival and the growth of online grocery are both readable as shifts in this variable and not in the retailers' strategies.

The evidence, and the criticism

The theory is descriptive and its status is contested in the same way as every cyclical account of institutional change.

The strongest criticism is that it is unfalsifiable as usually stated. Because assortment width is always either growing or shrinking somewhere, almost any observation can be fitted to the cycle after the event, and the theory makes no prediction about timing or amplitude. A model that explains every outcome equally well predicts none of them.

The second criticism is that it treats retail formats as the unit of analysis when the real movement often happens inside a single firm. A supermarket that has narrowed its electricals and widened its ready meals has moved in both directions at once, and a single measure of assortment width conceals that.

The third is contextual. Hollander's evidence is drawn from developed Western economies over a period of rising affluence and increasing mobility, and the sequence does not reproduce cleanly in markets where modern retail arrived all at once.

What survives the criticism is the observation itself. Assortment width does oscillate, the oscillation is driven by identifiable pressures, and a retailer choosing a position on that dimension is making a decision that has a predictable set of consequences and not a permanent identity.

What the theory implies for a retailer today

Read as a practical model rather than a history, the accordion carries three implications that a strategy answer can use.

Position is temporary and should be chosen deliberately. A retailer at either extreme is, on this account, already generating the pressure that will move it. Knowing which direction that pressure runs is more useful than defending the current position as an identity.

The dangerous place is the middle. A range too wide to carry authority in any category and too narrow to complete a shopping trip has the cost base of the generalist and the catchment of the specialist. Many high-street failures are legible this way: not a wrong choice of position, but the absence of one.

Online changes the catchment, not the logic. The constraint that made specialisation viable only above a certain population is largely gone, which is why extremely narrow specialists — a single product category, sometimes a single product — now exist where they could not have before. The accordion still operates; it operates over a much larger catchment, and correspondingly deeper ranges.

The current movement is towards the wide end in grocery, where marketplaces attach third-party ranges to a food business, and towards the narrow end in categories where subscription or direct-to-consumer models have found depth the generalists cannot match. Both are visible at the same time, which is exactly what the theory predicts and exactly why it predicts so little.

Using it alongside the wheel of retailing

The two Hollander theories are frequently confused, and distinguishing them is usually worth a mark.

The wheel of retailing describes movement on price and service: innovators enter as low-cost, low-margin, low-service operators, trade up over time as they add service and fixtures, and leave a gap at the bottom for the next entrant.

Accordion theory describes movement on assortment width, and says nothing about price.

The two are orthogonal and can be applied together. A discounter entering with a narrow range at low prices is at one point on both dimensions; as it trades up it may also widen, moving on both at once, and the analysis is stronger for separating the two movements rather than describing them as one trend.

A third companion, the dialectic process account, treats the outcome as a synthesis: the department store and the discount store combining into the discount department store. Where the accordion sees oscillation between two poles, the dialectic sees a new format containing elements of both, and on some histories that fits the evidence better.

Measuring assortment width, which the theory leaves vague

One practical weakness is worth raising because it affects any attempt to test the model: Hollander does not say how width should be counted.

Three measures are in use and they disagree. Counting categories treats a supermarket adding a pharmacy counter as widening, even if the pharmacy carries twenty lines. Counting stock-keeping units treats deepening an existing category as widening, which is the opposite of what the theory means. Counting share of floor space by category captures the trade-off a retailer actually faces but is unavailable for online businesses, where shelf space is not scarce.

For an online retailer the concept strains further. A marketplace listing millions of items across every category is maximally wide by any count, and yet the customer's experience of it is closer to a specialist, because search and not layout determines what they see. Whether that is a fourth turn of the cycle or evidence that the dimension has stopped describing anything useful is a fair question to raise, and raising it is worth more than applying the model as though the measurement problem did not exist.

How it is examined

Questions in this area normally supply a retail history and ask what explains it.

The workable approach is to plot the assortment width at each stage explicitly, in one sentence each, then name the pressure that caused each transition, not asserting that a cycle occurred. Where the case shows a firm widening in one category and narrowing in another, say so, because that is the observation the theory handles least well and noticing it demonstrates more than applying the model cleanly would.

Finish on the limitation that fits the case. If the history is short, the timing objection is the right one. If the movement is internal to one retailer, the unit-of-analysis objection is. Choosing the criticism that the evidence actually supports reads very differently from listing three.