Abhishek S.
Shipping in public. Listening in private.

Abhishek

I lead women’s Indo-Western & Premium at Max Fashion. I also wrote the AI that runs the buying floor.

Rare profile. Category operator who ships production code.

Senior Buying Leader · Max Fashion Women’s Indo-Western & Premium · 530+ India stores NIFT ’12 · Twelve years on the floor

abhishek@bengaluru ~ %
>role: senior buying lead
>dept: women’s indo-western + premium
>floor: 530+ stores india

Time-Based Competition

In 1988, George Stalk Jr. published a Harvard Business Review article arguing that the most dangerous Japanese manufacturers were not winning on cost or quality. They were winning on time. Honda was designing a new car in roughly half the calendar time Detroit needed. Toyota was retooling stamping presses in single-digit minutes against Western shifts of hours. The framework's claim was that compressing cycle time was not an operational nicety; it dragged price, quality, share, and inventory turns behind it as second-order effects.

The core claim

Stalk's argument inverted the conventional sequence. The Boston Consulting Group orthodoxy of the 1970s read: scale lowers cost, cost wins share, share funds scale. Time-based competition swapped the lead variable. Compress the cycle between order and delivery, or between concept and shelf, and three things follow without separate effort.

First, working capital falls because inventory sits for less time. Toyota's just-in-time discipline cut inventory turns from the Western norm of roughly 5 per year to over 70 in some plants by the mid-1980s. Second, forecast error stops mattering as much, because you are responding to demand instead of predicting it months ahead. Third, the customer pays a premium for responsiveness that competitors structurally cannot match. The faster firm sells at full price; the slower firm marks down what nobody wanted.

Stalk called the slow firms "time-disadvantaged" and observed they typically had three to four times the value-added employees per unit of output, because most workers were managing the consequences of slowness rather than producing.

Where the framework predicted dominance

The 1988 article named cycle-time leaders before they were household names outside their industries. The pattern repeated across sectors that looked unrelated.

Firm Conventional cycle Cycle compressed to Year benchmark holds
Toyota (vehicle assembly) ~18 hours ~13 hours 1986
Honda (new model design) 5 years ~3 years mid-1980s
Zara (concept to store) ~6 months industry ~3 weeks by late 1990s
TSMC (process-node ramp) several quarters weeks ahead of rivals 2010s onward
Dell (build to ship) weeks of channel inventory hours late 1990s

Zara is the cleanest validation. Inditex was a regional Spanish manufacturer when Stalk wrote; by 2005, Zara's three-week design-to-shelf cycle had built the largest apparel group in the world. The mechanism was exactly what the framework predicted: faster cycle meant less inventory risk, fewer markdowns, and higher full-price sell-through, which funded the operational investment that kept the cycle short. The flywheel is self-reinforcing once it spins.

TSMC is the second validation, in a sector with no obvious surface link to apparel. Process-node leadership in semiconductors is increasingly about ramp speed, not absolute capability. Being 6 months earlier to volume on a node lets you charge premium prices to the only customers who can pay them (Apple, NVIDIA). Slow ramps subsidize fast ones. See concept process node leadership.

What slows firms down — and why most cannot fix it

Stalk's diagnostic was that 95% of elapsed time in most processes is wait time. The work itself is a tiny fraction. Cycle-time compression is not about working faster; it is about removing the queues.

The reason most firms cannot copy Toyota or Zara is structural. Queues exist because batch sizes are large, because functional handoffs require approval, because forecasting drives production rather than demand. Each of these is defended by a department whose performance metric depends on the queue existing. Procurement is rewarded for volume discounts (large batches). Finance is rewarded for capacity utilization (full machines). Merchandising is rewarded for forecast accuracy (long lead times). The cycle is long because each department is locally rational.

The firms that compressed cycle time, almost without exception, reorganized around the flow rather than the function. This is the operational analog of concept conways law: your product's cycle time mirrors your org chart's handoff count.

What's contested

Two challenges to the framework deserve honest treatment.

The first is that time-based competition is sometimes confused with lean manufacturing, and the two are not identical. Lean is a set of practices (kanban, kaizen, 5S). Time-based competition is a strategic claim about which variable to optimize. A firm can be lean and still slow if the lean practices serve a long-batch business model.

The second is the question of whether the framework still holds in software-mediated industries. Some argue that DevOps and continuous deployment have absorbed the time-based logic so thoroughly that it no longer differentiates. Others argue the opposite: that AI-mediated design and supply chain orchestration are about to compress cycles by another order of magnitude, and the firms that ignore this will be displaced as completely as Detroit was. The empirical jury is out as of 2026.

Why this has to do with other realms

The clearest cross-realm bridge is to evolutionary biology. The faster a population cycles generations, the faster it adapts. E. coli divides every 20 minutes; elephants every 15 years. The mathematical structure of selection rewards cycle time, not size or strength in absolute terms. Stalk's framework is, in this sense, an application of concept r selection vs k selection to corporate strategy. Firms with shorter cycles sample the demand environment more often and discard losing variants faster. This is why fast-fashion is structurally hard to attack with slow-fashion incumbents, and why the same logic shows up in venture capital portfolio construction.

An open question

If AI-mediated design tools compress the concept-to-prototype cycle by another 10x in the next five years, does the bottleneck move to physical production, regulatory approval, or human attention? The first firms to figure out which constraint actually binds will set the next generation of the framework.

Key sources

Further reading

Abhishek's take

I see this in the buy plan before I see it in any report. When a 100-day lead time becomes the fixed object, the team starts defending guesses; when a shorter drop window is possible, the same team starts reading the floor. The decision changes from “how much do I believe this forecast?” to “how cheaply can I be wrong before the next cut?”

See Also