Feedback Loop

A feedback loop is any arrangement in which the result of a process is fed back into that same process as an input, so the system acts on itself. Instead of running in a straight line from cause to effect and stopping, the effect loops around and becomes part of the next cause. Feedback loops are everywhere. A thermostat reads the room temperature and adjusts the heater, which changes the temperature it will read next. Word of mouth turns satisfied customers into a source of new customers, who can become more sources still. The defining idea is circular causation — output influencing future output through the loop. That circularity is what gives feedback loops their power and their surprises, because a system that shapes its own inputs can behave very differently from one that does not.

Feedback loops come in two fundamental kinds, and telling them apart is the whole game. A reinforcing loop (also called positive feedback) amplifies change: more of something leads to even more of it, so the system accelerates away from where it started. A balancing loop (negative feedback) counteracts change: a deviation triggers a response that pushes back toward a target, so the system settles and stabilizes. 'Positive' and 'negative' here describe direction, not desirability — a reinforcing loop can be virtuous or vicious, and a balancing loop can hold something good or bad in place. Most real systems contain several loops at once, and their interaction — which loop dominates when — determines whether the system grows, collapses, oscillates, or holds steady.

A reinforcing feedback loop feeds on itself. Each turn of the loop makes the next turn bigger, producing exponential growth or, in the other direction, collapse. Compound interest is a reinforcing loop, because interest earns more interest. So is a referral engine, where new customers refer more customers, and so is a bank run, where withdrawals trigger more withdrawals. Reinforcing loops are the engines of both explosive success and runaway failure, because they have no built-in brake — they keep amplifying until some other force intervenes. This is why founders prize reinforcing loops in growth, from network effects to viral referrals, and why they fear them in churn or reputation, where a bad experience feeds negative word of mouth that feeds more bad outcomes.

A balancing feedback loop does the opposite. It resists change and seeks equilibrium. It has a goal — a target temperature, a desired inventory level, a market price — and whenever the system drifts away, the loop generates a correction that pushes it back. A thermostat is the classic example, and so is a market where high prices dampen demand and pull prices back down. Balancing loops create stability and self-regulation, but they also produce resistance to intended change. Push on a system governed by a strong balancing loop and it pushes back, which is why some improvement efforts mysteriously fade. The practical distinction is direction of effect. Reinforcing loops amplify and destabilize, while balancing loops dampen and stabilize. Read a system by asking, for each loop, whether it adds to a change or opposes it — that tells you where the growth, the limits, and the surprises will come from.

Thinking in feedback loops changes how you diagnose and design systems. When something grows or collapses faster than a linear story can explain, look for a reinforcing loop driving it, and ask what will eventually limit it, because reinforcing loops never run forever — a balancing loop always shows up to cap them. When something stubbornly resists your efforts to change it, look for a balancing loop defending the status quo. In growth work, the goal is often to build and strengthen reinforcing loops that turn customers into more customers — referrals, network effects, content that compounds — while adding balancing loops that keep the system from overheating or degrading. Designing the loops, not just pulling one-time levers, is what makes growth self-sustaining rather than dependent on constant new input.

The traps come from ignoring the loop structure. Treating a system as a straight line misses that its output shapes its input, so you are surprised when a small push produces a huge effect (a reinforcing loop) or when a hard push produces almost nothing (a balancing loop absorbing it). Confusing 'positive' with 'good' is another error, because a reinforcing loop can be a death spiral and a balancing loop can lock in mediocrity. Assuming a reinforcing loop will grow without limit ignores the balancing loops that inevitably constrain it. The discipline is to map the loops, identify which are reinforcing and which balancing, see which dominates and when, and intervene where the loop structure gives you leverage — strengthening the loops you want and weakening the ones you do not.

Feedback originated in electrical engineering around 1900 for routing part of an output back to the input; cybernetics and systems thinking later generalized the feedback loop to any self-influencing system.

Etymology: source.

Search interest for this term over the last five years:

View interest-over-time on Google Trends →

Areas of marketing where feedback loop is a core concern: