Feedback loop
Also called a feedback cycle.
A feedback loop is a process in which a system's results feed back into it and affect what happens next. In business, this might mean using customer behavior to change a product, then measuring how customers respond.
How it works
A business team can build a feedback loop through a repeated sequence:
Action → result → measurement → adjustment → new result.
Collecting comments alone does not close the loop. The feedback must influence what happens next.
There are two basic types:
- Reinforcing, or positive, feedback amplifies a change. More buyers attract more sellers, whose products attract more buyers.
- Balancing, or negative, feedback counteracts a change. An inventory system orders replacements when stock falls below its target.
"Positive" does not mean good, and "negative" does not mean bad. Either type can help or hurt a business.
Raising onboarding completion from 40% to 60%
A hypothetical startup tracks whether new users finish onboarding. Customer feedback points to a confusing setup screen, so the team simplifies it.
| Measure | Before the change | After the change |
|---|---|---|
| New signups | 200 | 200 |
| Completed onboarding | 80 | 120 |
| Completion rate | 80 ÷ 200 = 40% | 120 ÷ 200 = 60% |
The observed gain is 60% − 40% = 20 percentage points, or 40 more completions per 200 signups.
That does not prove the redesign caused the improvement: the groups might differ. A randomized test would provide stronger evidence. The loop continues when the team uses these results to decide what to keep or test next.
For a founder, the benefit is fewer product decisions based only on opinions.
Questions people ask
What is a feedback loop in AI?
An AI feedback loop occurs when outputs, or responses to those outputs, influence later training or decisions. A recommendation system might learn from clicks and show more similar content. Those recommendations then shape future clicks, potentially reinforcing preferences or errors.
