Feedforward vs feedback: when future-focus works
Feedforward skips the argument about what happened and asks for suggestions instead. Where that beats feedback, where it quietly fails, and how to use both.
On this page
Most feedback conversations die in the same place: the two of you end up litigating what actually happened. You say the handoff was unclear; they say the brief changed twice. Ten minutes gone, nothing learned, both slightly annoyed. Feedforward is the move that skips that argument entirely — instead of assessing the past, you ask for suggestions about the future. "I'm working on being clearer when I hand work over. What are two things I could do next time?" Nobody can dispute a suggestion about tomorrow the way they can dispute an account of yesterday, so the defensiveness never gets a foothold. That's a real advantage, and it's also why feedforward gets oversold: there's one specific job it cannot do at all.
What feedforward actually is
The idea comes from Marshall Goldsmith, whose work on behavioural change in executives built it into a repeatable exercise. The rules are deliberately strict, and the strictness is the point:
- You name the behaviour you're working on. Not "give me feedback" — a specific thing: interrupting less, delegating without taking it back, being clearer on priorities.
- They give suggestions for the future only. No assessment of how you've been doing, no examples of you getting it wrong. Only ideas for next time.
- You say thank you and nothing else. No defending, no explaining the context, no "yes but". Just take it.
- Anyone can play. Because it isn't a judgment of your history, it doesn't require someone who's watched you closely. A peer from another team can offer a useful suggestion about running clearer handoffs.
That last rule is the counterintuitive one. Feedback needs a witness; feedforward only needs someone with ideas.
Why it defuses the reaction feedback triggers
Feedback about the past is, structurally, a claim about who you have been — which is why it lands as a verdict and why the defensive reflex fires before anyone has thought about it. Amy Edmondson's research on voice and safety at work points at the same asymmetry from the other side: people engage far more readily when a conversation is framed as a shared problem to solve than as an evaluation to survive.
Feedforward moves the whole exchange into the second frame. There's no history to defend, so the receiver's status isn't on the line, and the giver doesn't have to nerve themselves up to deliver bad news. That's genuinely easier for both people — which matters, because the feedback that never happens is the most expensive kind.
The comparison, honestly
Feedback
Best for: establishing that something needs to change
Anchored to a specific past moment, so it can prove a problem exists and calibrate how you're actually landing. Disputable, threatening, harder to give — and the only tool that can tell someone something they don't already know about themselves.
Feedforward
Best for: generating ideas once you know what to change
Suggestions for the future only, so it's low-threat, easy to ask for, and available from anyone. But it assumes you already know which behaviour to work on — it cannot surface a blind spot, and it can't hold anyone accountable.
Where feedforward quietly fails
Three limits, and the first is the serious one.
It cannot surface a blind spot. Feedforward starts with you naming the behaviour. If you don't know that you cut people off in meetings, you will never ask for suggestions about it — so the thing most worth fixing is precisely the thing feedforward can't reach. It's a tool for closing a gap you've already identified, not for finding one. Any advice that presents feedforward as a replacement for feedback has quietly assumed the hardest part is already solved.
It can't establish accountability. If someone's work is genuinely below the bar, a stream of friendly suggestions about next time doesn't tell them that. At some point a manager has to say plainly what must change and by when — and that conversation is anchored in what actually happened. Feedforward is not a way around it; using it as one is just avoidance wearing a nicer face.
It stops working on a repeated pattern. The third time the same thing happens, more suggestions aren't the answer — the recurrence means your diagnosis was wrong, and that needs a different conversation entirely.
How it relates to SBI, COIN and STAR
This trips people up, because the models get listed side by side as if they were alternatives. They aren't the same kind of thing.
SBI, COIN and STAR are structures for delivering a past-anchored observation — they exist to keep you specific and behavioural instead of vague and personal. Pick between them based on the moment in front of you; which model fits when is its own decision.
Feedforward is a different act altogether. It isn't a way of phrasing an observation; it's a request for suggestions, with the observation deliberately removed. So "feedforward vs COIN" isn't really a choice between two models — it's a choice between telling someone how something landed and asking someone how to do it better next time.
Which means the useful question isn't which one wins. It's which order to use them in.
Use both: diagnose backwards, fix forwards
The combination is stronger than either alone, and the sequence matters.
Diagnose with feedback
Establish what's actually happening, anchored to specific recent moments — because this is the only step that can tell someone something they can't see. Use SBI or COIN here to stay behavioural.Agree on the one behaviour
Name the single observable thing to change. Feedforward needs a target, and a vague one ("communicate better") produces vague suggestions.Switch to feedforward for the fix
Now ask several people for two suggestions each about that behaviour, going forward. This is where the ideas come from, and where nobody has to defend anything.Say thank you — genuinely nothing else
Resist explaining why a suggestion won't work. The moment you start assessing the suggestions, you've turned a low-threat exchange back into a debate, and people stop offering.Re-measure with feedback
Suggestions don't tell you whether the change landed. Only an outside read does — so close the loop where you opened it, a quarter later, with the same questions you started from.
What this looks like in practice
A team lead has been told — via a structured round, not a hallway comment — that his direct reports rate "gives clear direction" far lower than he rates himself. That's the feedback half doing the thing only it can do: he genuinely didn't know. Arguing about which specific week was unclear would waste the insight, so he doesn't. He picks one behaviour (state the single priority at the start of each project, in writing) and then goes feedforward: he asks five people — two reports, two peers, and someone from another team who's never seen him run a project — for two suggestions each on making priorities land.
The outsider's suggestion turns out to be the most useful one: put the priority in the channel, not the kickoff meeting, so it's still there in week three. She had no idea how he'd been operating, which under feedback rules would have disqualified her, and under feedforward rules made no difference at all. He says thank you to all five, changes the one thing, and re-runs the same questions in October to find out whether it actually moved. That's the full loop: backwards to find it, forwards to fix it, backwards again to confirm.
Where Mirorly fits
Notice which half of that loop is hard to get. Suggestions are cheap — ask five people and you'll have ten. What's scarce is step one: an honest read of which behaviour to work on in the first place, especially the ones you can't see and nobody volunteers. Feedforward is genuinely good at closing a known gap and structurally incapable of finding one. Mirorly is the other half: you answer behavioural questions about how you work, your team answers the same ones with an anonymity threshold so honesty is safe, and you read your view next to theirs — so the behaviour you take into a feedforward round is the one that actually matters, not the one you happened to think of. The core leadership behaviours template is built for that, and running it again next quarter is how you find out the suggestions worked.
Common questions
The one-line summary
Feedforward removes the past from the conversation and asks only for suggestions about next time, which strips out the defensiveness that kills most feedback — but the same move makes it structurally unable to surface a blind spot or hold anyone accountable; so it isn't a replacement for feedback or an alternative to SBI and COIN, it's the second half of a loop that has to start with an honest read of what to change, and end with re-measuring whether it did.