How to Design the Systems That Run Your Business
The operator's guide to feedback loops
The School of Knowledge is the weekly newsletter for SME owners and investors who want frameworks they can actually use — frameworks, checklists, and operating manuals every weekend, built to read on Sunday and use on Monday.
In the big picture, one store’s inventory problem may seem trivial and fixable. But imagine that the inventory is that of all the unsold automobiles in America. Orders for more or fewer cars affect production not only at assembly plants and parts factories, but also at steel mills, rubber and glass plants, textile producers, and energy producers. Everywhere in this system are perception delays, production delays, delivery delays, and construction delays. Now consider the link between car production and jobs—increased production increases the number of jobs allowing more people to buy cars. That’s a reinforcing loop, which also works in the opposite direction: less production, fewer jobs, fewer car sales, less production. Put in another reinforcing loop, as speculators buy and sell shares in the auto and auto-supply companies based on their recent performance, so that an upsurge in production produces an upsurge in stock price, and vice versa. That very large system, with interconnected industries responding to each other through delays, entraining each other in their oscillations, and being amplified by multipliers and speculators, is the primary cause of business cycles.
The difference between a systems thinker and a system builder lies in a single verb: design. To read Meadows is to learn how to see feedback loops in existing systems—to trace reinforcing spirals and identify balancing mechanisms. But the operator’s question is harder: how do you engineer feedback into a system that does not yet have it? How do you decide what signal to amplify, what delay to shorten, what correction to automate? The shift from observation to architecture is the shift from analysis to craft, and it is the foundational skill of anyone who builds organisations, products, or supply chains.
A well-designed feedback loop compounds because every major decision is evaluated against the loop’s logic, not against isolated metrics.
Zara’s customers shop at their stores 250% more than other fashion retailers—not because their clothes are inherently better than everybody else’s—but because they understood how to design effective feedback loops.
The flagship brand of Inditex, Zara’s feedback loop was built on speed and efficiency. Traditional fashion retailers typically take six to nine months to go from design to shops, with most of production done pre-season. Zara has compressed this down to fifteen days. Under a quarter is designed before the season begins, with over half designed, manufactured, and shipped mid-season in direct response to customer demand. Store managers across more than six thousand shops in over eighty countries transmit sales data and qualitative observations—”this colour shirt isn’t selling,” “we can’t ever have enough stock of this winter coat”—to a centralised data-processing centre that operates around the clock. Small initial batches, sometimes just a few items, are sent out to test demand and feed data back into the system.
The logic behind the loop runs as follows: fast response to demand generates higher sell-through rates (roughly 85 per cent of items sell at full price, against an industry average of around 60 per cent), which generates better margins, which funds the proximity manufacturing infrastructure. There was also a deliberate choice to house roughly 50 per cent of production hubs in Spain, Portugal, Morocco and Turkey in lieu of Asia to make this kind of speed possible in the first place. The average Zara customer visits seventeen times per year, compared with three or four for a typical retailer, because the product mix changes constantly.
The feedback loop is not merely observed; it is architecturally embedded in every operational choice—from the location of factories, to the design of the RFID-tagged supply chain, to the twice-weekly ordering cadence.
But not every feedback loop requires speed to be effective—and not every successful feedback loop has to be reinforcing.
Between 1948 and 1975, Taiichi Ohno and Eiji Toyoda created Toyota’s Andon system—one of the most elegant balancing feedback loops ever designed for a production environment. When a worker on the assembly line spots a fault, defect, shortage, or safety hazard, they pull a cord (or press a button in modern versions). A coloured light displays above the assembly line—green for normal, yellow for help requested, and red for line stopped. The word “Andon” means “paper lantern” in Japanese, hence the colours. Once initiated, a team leader goes over to thank the worker who pulled the cord. If the problem can be fixed within the cycle time, the line continues; if not, the affected station is stopped until the problem is resolved. The frequency and nature of every stop is logged and fed into Toyota’s continuous improvement process, Kaizen.
Problems on a Western production line are thought of as poor performance, whereas at Toyota—they are seen as the highest form of quality control. At one Toyota plant, when the average number of Andon pulls per shift dropped from a thousand to seven hundred, the CEO called an all-hands meeting—not to celebrate but to investigate. Either problems were being hidden or standards were too low. For Toyota, the absence of correction is itself a danger signal because it implies the balancing feedback loop has gone quiet.
The NUMMI joint venture between Toyota and GM in the 1980s proved the system was transferable: the Fremont, California workforce, previously considered GM’s worst, became its best within a year of adopting the Andon system and its cultural infrastructure.
Toyota embedded in their production lines a system that rewarded patience over speed, quality over output metrics. They demonstrated another clear understanding of feedback design when they also created a negative feedback loop—with the purpose of mitigating material waste.
Toyota’s pull-based system ensures materials are only moved down the production line as and when they are needed, resulting in maximum efficiency. This is the inverse of a push-based system that forecasts when certain materials are expected to be needed.
At Toyota, Kanban cards are physical tokens of demand, visible to all. The card-based production control system operates as a balancing feedback loop: downstream workstations signal to upstream workstations when they need more material. This pull-based system creates a negative feedback loop that regulates production by actual consumption rather than by forecast, preventing the overproduction that bedevils push-based systems.
The Zara and Toyota cases are instructive, but they're the easy half of this essay. The harder question is what happens when feedback loops go wrong — and more usefully, what you actually do about it. Below i've covered three failure modes (with the case studies to match), the single highest-leverage intervention for each, and a practical audit framework you can apply to your own business. If you've ever explained a bad quarter with "the market shifted" or "we didn't see it coming," that section is probably worth your time. There's also a downloadable worksheet for paid subscribers.
When feedback loops fail: too slow, too fast, or miscalibrated
