What Is a Feedback Loop? Examples, Types & How It Works

A feedback loop is a circuit in which the output of a process is measured and fed back as input to the next round of the same process, so the system adjusts itself. In customer and product work the circuit runs from someone raising something, through a decision and a change, back to that person being told what happened.

The loop is only closed when the last step happens. Everything up to it is collection.

Open Loops vs Closed Loops

This distinction is the whole subject. An open loop takes input and never returns anything to the source. A closed loop returns.

	Open loop	Closed loop
Input	Collected	Collected
Decision	May happen	Happens and is recorded
Change	May ship	Ships, or is declined explicitly
The submitter learns	Nothing	What was decided and why
Next round of input	Declines, quietly	Sustains or grows
What it feels like from outside	Shouting into a form	A conversation
Failure is visible	No	Yes

Most feedback programmes are open loops that believe they are closed, because the team can see the decisions internally. The submitter cannot. From outside, a decision nobody was told about and a decision nobody made look identical.

The expensive consequence is not the unhappy user. It is that an open loop degrades its own input: the people who stopped hearing back stop writing, and what remains is a smaller, more self-selected sample that the team keeps treating as representative.

The Six Stages

1. Collect  ->  2. Understand  ->  3. Decide  ->  4. Act  ->  5. Communicate  ->  6. Measure
                                                                                       |
                                                                                       v
                                                                          back into stage 1

Collect. Input arrives through channels you chose.

Understand. Group, deduplicate and interpret. A request is usually someone's proposed solution to a problem they have not described, and the problem is the useful part.

Decide. Including deciding not to. A recorded "no, because" is a closed loop; silence is not.

Act. Ship the change, or make the alternative real.

Communicate. Tell the people who raised it. This step converts an open loop into a closed one, and it is the step under time pressure.

Measure. Did the change do what it was supposed to. This is the stage that makes it a loop rather than a line, and it is skipped more often than communication is.

Stages 5 and 6 are the two that get dropped, and dropping either leaves you with a pipeline, not a loop.

Positive and Negative Feedback Loops

These terms are borrowed from systems theory, where they mean something different from what most people assume, and the confusion is worth clearing up because the words appear constantly in product writing.

A negative feedback loop is self-correcting. The output opposes the deviation and the system stabilises. A thermostat is the standard example: too cold, heat on, temperature rises, heat off. Most customer feedback loops are meant to be negative in this sense. Complaints rise, the team fixes the cause, complaints fall.

A positive feedback loop is self-reinforcing. The output amplifies the deviation and the system runs away from where it started, in either direction. Referrals producing more users who produce more referrals is a positive loop. So is a support backlog: slower replies produce more chasing messages, which produce slower replies.

Neither term describes sentiment. "Positive feedback" in the everyday sense means praise; a positive feedback loop can be a collapse. Pointing the pattern at retention specifically means starting from churn signals rather than submissions. Teams that conflate the two end up describing a runaway churn spiral as a negative loop because the news is bad, which reverses the diagnosis.

A Complete Example

A small team ships a billing product. In one month, four users write in to say invoice PDFs are missing the purchase order number their finance departments require.

Collect: three arrive through support, one on the suggestion board.

Understand: all four are the same request under different wording, so they are merged. The underlying problem is not "add a PO field", it is "our invoices get rejected by our customers' accounts payable teams", which turns out to also affect the invoice email subject line.

Decide: the team commits to the PO field and declines the subject-line change for now, recording why.

Act: the field ships two weeks later.

Communicate: the four submitters are told, by reply and through the changelog entry, including the explicit "not yet" on the second part.

Measure: the team checks whether invoice-related support volume falls over the following two months, and whether the declined item comes back.

That last stage is what separates a loop from a queue. Without it the team has shipped a feature and learned nothing about whether it solved the problem.

Internal and External Loops

Not every loop involves a customer. Four run in most product organisations, and they fail in different ways:

Customer to product. The one everyone means. Fails at communication.
Support to product. Support sees repeated breakage first. Fails when there is no route from a ticket trend into planning.
Sales to product. Sales hears objections and unmet needs. Fails by over-weighting the deal in front of them.
Product to customer. Roadmaps, changelogs, release notes. Fails by going quiet, which reads externally as the product being abandoned.

A team can have a healthy customer loop and a broken support loop, and the symptom is the same request arriving as new every quarter.

Why Feedback Loops Break

Scattered collection. Items live in the tool they arrived in and are never counted together.
No owner. Everyone can read the queue, so nobody is responsible for it.
No decision step. Items are read and left open, which is a decision made by expiry.
No status visible to the submitter. The work happens and looks like nothing from outside.
Communication treated as optional. The step that closes the loop is the one with no internal deadline.
No measurement. Nobody checks whether the change worked, so the same problem returns rescoped.
Cycle time longer than memory. A reply nine months after the request reads as a form letter regardless of how sincere it is.

Feedback Loop vs Continuous Feedback Loop

A feedback loop is the pattern: input arrives, someone decides, something changes, and the person who spoke is told. It describes one circuit and it can run once.

A continuous feedback loop is the operating model: the same circuit running permanently, with no start and no end date, so intake and response both stay open.

The difference shows up in scheduling. A loop can be a project: run a survey, read the results, ship a change, email the respondents, done. A continuous loop cannot, because there is no point at which you have finished, which means it needs an owner, a cadence and a queue that does not depend on someone remembering. Plenty of teams have loops and no continuous loop: feedback arrives all year and gets answered in bursts before planning. Closing the loop with automation is how the response side keeps up with intake that never pauses.

Measuring Loop Health

Use counts you already hold rather than imported benchmarks:

Cycle time from submission to a recorded decision, and separately to the submitter being told.
Closure rate: the share of items that reached a recorded outcome, declines included.
Communication rate: the share of closed items where the submitter was actually notified.
Reopen or recurrence rate: how often a solved theme comes back.
Input trend per channel, watching for a channel going quiet, which usually means the loop stopped closing there.
Post-change measurement rate: how often stage 6 actually happened.

Compare each against your own previous period. A loop is healthy when it is getting faster and staying closed, not when it hits somebody else's number.

Feedback Loops and Retention

The honest version of the relationship is narrower than it is usually stated. A feedback loop does not reduce churn by existing. What it can do is shorten the distance between a problem appearing and the affected customer knowing it is being handled, and give the team an earlier, better-corroborated view of which problems are costing accounts.

Whether that translates into retention depends on what the team does with it. A loop that closes quickly on requests nobody cared much about will not save an account leaving over something else. Cancellation and support signals feed the loop precisely because they identify the problems that carry revenue, and churn signals in support tickets is a better starting point than volume.

Tools

At low volume the mechanism can be a shared inbox, a spreadsheet and a discipline of always replying. What breaks first as volume grows is not collection but the closing step, because it requires knowing who asked for what, months later. That is the capability worth buying: a record linking every submitter to the item, and a surface that tells them when it moves.

FlagUp collects through widgets and boards, groups duplicates so a theme keeps all its submitters attached, and publishes a roadmap and changelog so stage 5 happens without assembling a mailing list by hand. See what each plan includes.

Frequently Asked Questions

What is a feedback loop in simple terms?

A cycle where what you learn from people changes what you do, and what you do gets reported back to them, so the next round of input is better informed than the last.

What is the difference between an open and a closed feedback loop?

A closed loop tells the person who gave the feedback what happened. An open loop does not. Everything else can be identical, and the open one still stops receiving input over time.

What are the stages of a feedback loop?

Collect, understand, decide, act, communicate, measure. Communication is what closes it and measurement is what makes it a loop rather than a one-way pipeline.

Are positive feedback loops good and negative ones bad?

No. In systems terms a negative loop is self-correcting and a positive loop is self-reinforcing, in whichever direction it starts. A support backlog that grows because replies are slow is a positive feedback loop with a bad outcome.

Does a feedback loop reduce churn?

Not on its own. It shortens the time between a problem appearing and it being handled, and it gives you evidence about which problems are attached to revenue. Whether retention improves depends on which problems you then choose to solve.

How long should a feedback loop take to close?

Faster than the submitter's memory of writing it. In practice teams do well to acknowledge within days, decide within a planning cycle, and report the outcome whenever it lands, including when the outcome is no.

FlagUp helps teams collect feedback in one place, decide what to build next, and tell the people who asked. Start free.