Feedback Loop: How Results Shape the Next Move
A feedback loop is a process in which the result of an action or change returns to influence the process that produced it. In MIT’s systems-thinking introduction, a decision affects the system, and the resulting state affects the same decision process later.
What makes feedback a loop?
The defining feature is a return path: an effect comes back to influence its own cause, rather than ending at an output, as illustrated in MIT’s explanation of closed-loop thinking.
The basic relationship is:
Action or change → result → feedback to the original process → further action or change.
In a decision process, observing results becomes part of a loop when those results inform a subsequent decision; MIT contrasts this return connection with a process that ends at implementation.
Positive and negative feedback loops
Systems thinking distinguishes positive feedback, also called reinforcing feedback, from negative feedback, also called balancing feedback; Donella Meadows describes the former as self-reinforcing and the latter as self-correcting.
| Type | How the return effect works | Resulting tendency |
|---|---|---|
| Positive or reinforcing | The response amplifies the original change, following Meadows’s description of positive feedback. | Further movement in the same direction |
| Negative or balancing | The response counteracts a deviation from a goal, following Meadows’s description of negative feedback. | Movement toward the goal |
The labels describe the direction of the effect, rather than whether it is desirable: NASA identifies a positive feedback that accelerates summer sea ice loss and a negative feedback that slows the decline.
Real examples of feedback loops
Thermostat: balancing room temperature
A thermostat regulates room temperature through a target setting, a device that monitors temperature, and a heating or cooling response, the three elements in Meadows’s thermostat example.
For heating, the loop can be written as temperature below the target → heating response → temperature rises toward the target → reduced need for heating, a balancing relationship illustrated by Meadows’s account of thermostat control.
Body temperature: negative feedback in biology
A biological negative feedback system has a sensor that detects a condition, a control center that compares it with the normal range, and an effector that produces a corrective response, according to OpenStax’s explanation of homeostasis.
In OpenStax’s temperature-regulation example, increased sweating helps remove heat when body temperature rises, while shivering produces heat during sufficiently cold conditions; these responses oppose the temperature change.
Childbirth: positive feedback with an endpoint
During labor, cervical stretching triggers signals that lead to oxytocin release; oxytocin strengthens uterine contractions, causing more stretching, as described in OpenStax’s childbirth example.
This is reinforcing feedback because the response strengthens the original stimulus, and OpenStax explains that birth ends the cycle by removing the cervical stretching that drives it.
Arctic sea ice: competing loops in one system
Summer sea ice melt and increased solar absorption create a reinforcing relationship in which melting contributes to conditions that promote further melting, the positive loop identified in NASA’s account of Arctic sea ice feedback.
The same system also contains a balancing effect: NASA’s 2018 study found that faster winter ice growth partially counteracted the decline without overcoming increased melting.
This documented combination shows why identifying one negative loop does not establish that an entire system is stable: NASA found that the summer reinforcing effect remained dominant in the study.
What is a customer feedback loop?
A customer feedback loop connects customer comments with a response and changes to the product or service, according to Qualtrics’s definition.
The process described by Qualtrics includes:
- Receive feedback, whether requested or unsolicited.
- Gather details and analyze what happened.
- Decide on the response to the customer and any internal action.
- Act on the findings.
- Communicate the planned or completed action to customers and employees.
Here, closing the loop means addressing the feedback and communicating the response, as explained in Qualtrics’s account of customer follow-up.
Customer-experience writing also uses positive and negative to describe favorable and unfavorable opinions, as Qualtrics does in its examples; those labels alone do not identify the reinforcing or balancing relationship defined above.
How to close a feedback loop in a decision process
For decisions, closing the loop means using observations to guide the next action and continuing to learn from its results, the iterative approach in the UK government’s Test and Learn guidance.
Its guidance on testing and iteration supports these steps:
- Define the question. Identify the outcome or assumption that needs examination.
- Set decision criteria. Specify the evidence that would justify continuing, adapting, or stopping before reviewing results.
- Collect suitable evidence. Choose observations, interviews, data analysis, or experiments in proportion to the question and its risks.
- Use the findings. Adapt the action or underlying assumptions, then gather feedback in the next iteration.
A feedback loop and a demonstration of causation are different achievements: the guidance distinguishes early signals used for iteration from the more robust evaluation needed to assess overall effectiveness and impact.
Why feedback timing matters
Feedback does not necessarily arrive immediately, and a decision can affect processes beyond the one being changed, both points emphasized in MIT’s discussion of delays and wider system effects.
A balancing response can also overshoot or oscillate when information or action arrives late, as Meadows explains in her discussion of feedback delays.
For evaluation, an early observation may therefore support a narrower decision than a long-term outcome would; Test and Learn distinguishes emerging short-term signals from evidence about longer-term effects.