Cited from real sources 6 min read Written by Arun Agrahri Updated October 2026

A framework by Marshall Goldsmith

Marshall Goldsmith's Feedforward: Ask for Ideas About the Future, Not Feedback About the Past

Feedforward is Marshall Goldsmith's alternative to feedback: you ask colleagues for ideas for the future instead of comments on the past. Each person gives one or two ideas, and your only reply is thank you. Goldsmith coaches CEOs and uses it to change one behavior at a time. It works only if you follow up every two months.

The framework

No feedback about the past, and no judging the ideas

Goldsmith defines the difference in one line:

one thing I teach is feed forward, which is different than feedback. Feedback is let's talk about your problems in the past. Feed forward is ideas for the future.

Goldsmith on feedforward versus feedback Watch at 22:32

The method has two rules. Rule one: no feedback about the past, because nobody can change it. Rule two is the one he calls hard: the person asking cannot judge the ideas they get. When he ran the exercise at Google in 2007, in a talk on his book What Got You Here Won't Get You There, he put it this way:

when people give you ideas don't say good idea bad idea I already knew that that'll never work no matter what people tell you you must stand there shut up listen

Goldsmith at Talks at Google Watch at 22:27

The only allowed reply is thank you. In a 2026 interview, a host answered one of his ideas with an explanation, and Goldsmith stopped him. He had set the rule earlier in the same conversation: "when I give advice, just say thank you. You don't say but."

Feedforward does not replace feedback in his coaching. Every CEO he works with first gets confidential feedback from colleagues, from 18 people on average. That feedback picks the behavior, and feedforward supplies the ideas for changing it.

How to apply it

How do you run feedforward with your team?

Steps 1 to 3 happen once. Steps 4 and 5 repeat every two months until colleagues see the change.

1

Pick one behavior.

Choose one thing, such as listening or being less impatient. Goldsmith's instruction to a room of Google employees: pick one area to improve, talk to people, follow up, and stick with it.
2

Get the diagnosis first.

Ask colleagues for confidential feedback, then confirm with your manager or board that you picked the right behavior. Do not start with a guess about your own weakness.
3

Thank them, apologize, and ask for ideas.

Thank everyone who gave feedback. Apologize for the behavior with no excuses. Then tell them what comes next.
4

Listen, take notes, and say thank you.

Do not judge, debate or explain. You do not have to use every idea. Goldsmith's clients tell colleagues they cannot promise to change everything, only to listen and do what they can.
5

Follow up every two months.

Go back to the same people. Remind them what you said you would work on, and ask for ideas for the next two months based on the last two. Keep going for a year.

On step 3, this is the sentence Goldsmith teaches his clients to say:

i'm not going to ask you for more feedback about the past i'm going to ask you for ideas for the future

Goldsmith on The Knowledge Project Watch at 5:36

For a whole team, have each member ask the others how they can better help the team in the future. When you give ideas to a very successful person, Goldsmith suggests telling them to use what works and ignore the rest. In his words: "If you can only use two of the ideas, you are still two ahead."

The founder version: a delegation dialogue

A founder who has become the bottleneck can use the same method on delegation. Goldsmith's 2025 version is a one-on-one with each direct report about their job:

Don't ask for feedback. Just in the future, feed forward. What are some areas where maybe I can be less involved that you think would be good?

Goldsmith on delegation for founders Watch at 37:01

Then ask two more questions. Where do you need more of me? And do you see me making decisions at my level that someone else could make? He says founders tend to do too much of what they like doing, which has nothing to do with what the business needs. In a February 2026 interview he gave the same three questions and added that a leader should look for things not to do.

When it works

When does feedforward fail?

Works best when

The person chose the behavior and wants to change it.
The person works on one behavior at a time.
Confidential feedback already named the problem.
The person goes back to the same colleagues every two months.

Fails when

The person does not care. Goldsmith: if people don't care, don't waste your time as a coach.
The person argues with the ideas or explains them away.
Nobody follows up.
Nobody ran the diagnosis first.

Skipping the follow-up is the failure Goldsmith measured. From his study of 86,000 people:

leaders that don't follow up and don't talk to people guess what don't get better improvement looks like random chance

Goldsmith on the 86,000-person study Watch at 7:22

He tells the same story about Alan Mulally, the former Ford CEO he coached. Mulally got confidential feedback and talked about it in public. Then he followed up over and over with feedforward.

For the opposite view on when to talk about the past, compare Kim Scott's Radical Candor, which asks for criticism given in person and close to the event. Goldsmith's habit of asking for help also needs humility, which Jim Collins puts at the center of Level 5 Leadership.

Sources

Where Goldsmith discusses this

We read five long Goldsmith recordings for this page on 2026-10-06, from a 2007 talk at Google to two interviews from February 2026. Three of them use the word feedforward. The other two walk through the same steps.