The Soliciting Feedback Playbook
Free. No email required.
Soliciting feedback well is its own skill, separate from giving it and separate from receiving it once it arrives. Waiting for feedback to show up on its own means settling for whatever someone else decides to volunteer. Asking is what closes that gap.
Giving feedback is its own playbook: see The Giving Feedback Playbook. Receiving it once it lands, planned or unplanned, solicited or not, is its own playbook too: see The Receiving Feedback Playbook.
One boundary worth naming: formal performance ratings and calibration are a different scope. This playbook covers informal, in-the-moment feedback only.
1. The 60-Second Version
“Almost no one I know ever requests or solicits feedback.” That's my own read on where the real gap is. Built-in processes are supposed to cover it, the annual review, the 360, and asked directly whether they do, my answer was “really not.” Soliciting reaches what giving and receiving can't reach on their own: without it, the only feedback you get is what someone else decides to volunteer. Choose who you ask on purpose. Ask about one named, specific thing, never “how am I doing.” Listen without defending or explaining. Say thank you, and decide for yourself what to do with what you get.
2. The Default Play
“Got any feedback for me?” mostly returns nothing, because most people haven't thought about what they'd say and don't know how to say it well on the spot. Two things fix that: who you ask, and what you ask them.
Choose providers on purpose: someone close enough to the work to have actually seen something real, and someone who'll tell you the truth rather than soften it into nothing. There's a third way this goes wrong worth naming on its own, since it isn't the same failure as softening: someone who'll give it to you straight, but only ever the harsh, unusable kind, nothing you can actually act on. Too-friendly, won't-give, only-unusable-negative all exist, and picking around all three is part of choosing well. It helps to say why you're asking, since giving honest feedback is uncomfortable for the other person too, and knowing your intent makes it safer for them to answer straight.
Then point the request at something specific and named. Bring the attempt you already made where you can: “I'm working on being more concise in meetings. Here's what I tried today. How did it land?” A start/stop/continue question, one at a time, keeps each ask small enough to actually answer.
Here's the ask itself, in six moves, my own version of it, built out of primary sources I've read and listened to for years: Tom Henschel's Look & Sound of Leadership podcast, the Radical Candor blog, and Lyssa Adkins' Coaching Agile Teams among them.
- Choose your providers deliberately. Someone close enough to the work to have actually seen something real, who'll tell you the truth rather than soften it.
- Name your intent before you ask. Say why you're asking, and say what it's not. Marshall Goldsmith calls this advertising your intent to change; Kim Scott's version is laying out the idea before you ask anyone to use it with you.
- Ask something specific, and keep it bounded on purpose. Start, stop, continue is my default shape for it; plus, minus, delta is the same technique, just my own name for it. The cap goes back further than either label, to a psychology professor named Phil Daniels, by way of Thomas DeLong's 2011 Harvard Business Review piece. Whether you run all three questions or just one depends on the moment; most of the time it's one.
- Listen, and say thank you. Brace for it, stay quiet, and listen to understand rather than to respond. Goldsmith's rule for feedforward is the strictest version of the same discipline: thank you, and nothing else, not even a compliment on the suggestion.
- You decide what to act on. Nothing here obligates you to act on everything you're handed; that choice, and the freedom to set something aside, stays yours.
- Go back after you act. Prioritize what you act on, then return to whoever gave it to you, tell them what changed, and ask for more. What that's actually looked like for me is below.
A few things make a question like that actually land, and the same shapes show up independently across sources that never cite each other. Bound the size of the answer on purpose: Stone and Heen's “name one thing,” a start/stop/continue variant that caps each list at three items, and Marshall Goldsmith's rule of exactly two suggestions all converge on the same move. An unbounded “what do you think” leaves too much room to say nothing at all; naming exactly what you want an answer about is what forces a real one back. Kim Scott's own go-to question, offered on the Radical Candor blog as a template to adapt rather than copy, adds two more rules worth borrowing: never ask anything answerable with a yes or a reflexive “everything's fine,” and make sure it sounds like something you'd actually say.
Goldsmith's feedforward is a second full technique, distinct enough to run alongside start/stop/continue rather than replace it. Pick one behavior to change. Describe it to someone in a one-on-one conversation. Ask for exactly two suggestions for the future, nothing about the past. Listen without judging a single one of them; the only permitted response is “thank you.” The rule that makes it different from everything else here: no mention of the past, ever.
Two more shapes are worth having on hand. Ask the identical, specific question of several named people and look for where the answers converge, rather than triangulating off one conversation, what Stone and Heen call listening for themes. And a different angle from “what should I improve”: ask what matters to the other person instead, “what's one thing I could change that would make a big difference to you,” which removes their worry about presuming to know what your growth needs.
The evidence backs the shape more specifically than it might look. A field study of 387 managers found that asking for negative feedback increased how accurately they understood where they actually stood, and raised how effective their own superiors, subordinates, and peers rated them; asking for positive feedback did the opposite (Ashford & Tsui, 1991). Stop and start are easier to answer than “how am I doing,” and they're the half of the question that pays off.
My own practice, day to day, hasn't been a single standing habit, and it's worth being honest about that rather than implying otherwise. I've tried direct messages, a written survey, team-wide asks, and raising it inside a 1-1, at different points, not on any fixed schedule. Historically it's looked like quarterly, or twice a year, or at least once a year, or asking when a project was ending, and moving from project to project as a consultant breaks whatever standing practice I'd otherwise build. It was genuinely easier to ask when I was an employee somewhere than it's been moving between engagements. The written survey is the version I've kept coming back to: carefully written, repeatedly rewritten questions, and I'm still iterating on it.
Running that at team scale isn't automatically safer than asking one person, and the guardrails are worth naming rather than skipping. A large meta-analysis of feedback research is widely cited for finding that more than a third of feedback interventions actually decrease performance (Kluger & DeNisi, 1996), and Marshall Goldsmith's own follow-up research, across 86,000 participants, led him to publicly reverse his stance on 360-style processes that don't have real follow-up behind them. A survey works with genuine safety behind it, kept separate from evaluation, and only if following up becomes a real discipline rather than a hope. Building the kind of team that tells you the truth in the first place, at scale, is its own separate practice: see “Surround Yourself with Courageous Truth-Tellers, Not Sycophantic ‘Yes Men’” rather than this playbook for that side of it.
Soliciting from your own manager deserves its own note, starting with the fear underneath it: whether asking itself reads as angling for something, rather than a genuine ask. My own answer is that it usually doesn't, but it could, so the ask needs managing on purpose: frame it plainly as improvement-seeking, not fishing. “I am trying to improve. I'm looking for feedback: things I do well, things I could improve on.” Said that way, it's hard to mistake for anything but a straight ask.
It also comes up short for a plainer reason, separate from intent. Managers usually have less day-to-day visibility into your actual work than the peers next to you do, and what moves up a hierarchy tends to get edited on the way, a pattern researchers call the hierarchical mum effect: subordinates soften or withhold information going upward to protect their own image, and the single most avoided topic in that editing is a supervisor's or peer's own performance (Bisel, Messersmith & Kelley, 2012; Milliken, Morrison & Hewlin, 2003). Two moves help. Ask permission before diving in, “Would it be helpful if I told you what I thought of X?”, and offer your own review before you ask for theirs, so it feels like a trade, not you sitting in judgment of them (Kim Scott). From there, the same discipline as everywhere else in this section applies, aimed upward: ask about something specific and recent, and if the first answer is a brush-off, a second, narrower question usually gets more.
Going around the chain to get ground-truth from below doesn't fix the same problem from a leader's side. My published position on skip-levels is that they tend to undermine the primary manager relationship more than they surface anything genuinely new. The better version of the same instinct: coach managers to run good 1-1s so the information surfaces where it should, and run skip-level meetings with the whole team rather than one person at a time.
3. Scripts
Start/stop/continue, one question at a time, is the default here: it's what these scripts lead with. Reach for feedforward instead when a past-focused ask keeps stalling out.
Start / stop / continue:
“What's one thing you'd like to see me start doing in my coding and testing work?”
“What's something I should do less of during team events?”
“What's something you want me to keep doing when it comes to story writing?”
Never “Do you have any feedback for me?” or “How am I doing?” Both return nothing usable.
The one-question version (Stone and Heen):
“What do you see me doing, or failing to do, that's getting in my own way?”
Impact, not improvement (Stone and Heen):
“What's one thing I could change that would make a big difference to you?”
A go-to question to adapt, not copy (Kim Scott's own template):
“What could I do or stop doing that would make it easier to work with me?”
Feedforward (Marshall Goldsmith), future only, never the past:
“Here's one thing I'm trying to get better at: being more concise in meetings. Two suggestions for what I could try next time, nothing about how it's gone before. ‘Thank you’ is all I'll say back.”
Naming your intention, up front:
“I'm asking because I want to get better at this, not to make myself feel good and not to put you on the spot. Just tell me straight.”
Managing the ask upward, so it doesn't read as fishing:
“I am trying to improve. I'm looking for feedback: things I do well, things I could improve on.”
Responding to what comes back:
“Thank you, that's useful.” Or, one clarifying question only: “Can you help me understand what that looked like?” Then thank you. Never explain, defend, challenge, argue, or cry.
If the first answer is a brush-off, don't fill the silence yourself. Count to six or seven and let it sit before you ask again.
4. The Follow-Through Loop
Getting feedback doesn't end when someone answers you. What happens after is its own process, the same way it is on the giving and receiving sides.
Here's what that's actually looked like for me. After a team survey, I recorded what came back in my own feedback profile, a running personal record, and shared it with my manager, where it fed directly into my performance review. For every constructive item, I wrote something down: why I think I got that feedback, what it would take to actually apply or learn from it, and how I'd change things going forward. As a consultant, moving between engagements, I can't always apply feedback as directly as an employee can inside one stable role, so the goal shifts to learning the pattern in it, positive or constructive, and thinking about how to apply that pattern going forward, even when the exact situation that produced it won't come back around.
The base mechanics underneath that come from the ask model above: you're not obligated to act on everything you're given. Prioritize what you do act on. After acting on something, go back to the person who gave it to you to talk it through and ask for more. Repeat, repeat, repeat.
One more thing worth building in on purpose: reward the first real piece of honest feedback out of proportion to its size. Kim Scott's account of a team in Tokyo and the Toyota red-box tradition both make the same point from different directions. A small, low-stakes first answer, thanked visibly and out of proportion to how small it was, is what makes the next, bigger piece of honesty feel safe to give.
5. Failure Modes and Fixes
The ask returns silence, or what comes back is too vague to use. This usually means the person doesn't yet trust that it's safe to answer honestly. Fix: go back to explaining your intention before asking again, smaller and more specific than the first try, and if it's still a brush-off, let the silence sit before you ask a second time rather than filling it yourself.
The same person keeps declining, or keeps giving you something too vague to use even after a second, narrower try. My own rule: retry once, differently. If you've asked two or three times and nothing usable comes back, that person probably isn't the right source for this; move on rather than pushing a fourth time. The same pattern run in reverse is worth noticing too: if several different people all come back blank on the same specific thing you asked about, that's worth re-evaluating on its own terms, whether the area is really worth worrying about, or whether other people genuinely don't see it the way you do.
Contradictory feedback from different people. One person says start, another says stop the same thing. Fix: ask the identical question of several named people and look for where the answers converge, rather than averaging two contradictory data points or triangulating off one extra opinion.
Won't asking make me look weak. The honest research answer runs close to the opposite: people who seek advice are seen as more competent, not less (Brooks, Gino & Schweitzer, 2015), and asking for feedback is generally associated with an enhanced image, not a damaged one (Ashford & Northcraft, 1992), with one exception worth naming rather than hiding: someone with an already-shaky performance track record doesn't get the same boost from asking.
Over-asking wears out the person you're asking. How often someone gets asked for feedback predicts their own stress, through the extra role it adds on top of their actual job (Krasman, 2018). Worth a real cadence, not just a good question; don't lean on the same one or two people every time.
Point an agent at this section to help you plan your ask: describe who you want feedback from and what you're working on, and have it draft a specific, named-focus question you can actually send.
6. AI-Era
Nothing here is specific to soliciting the way the follow-through loop above is; this ground is shared with the other two playbooks in this family. Rehearsing before a hard ask works the same way rehearsing before a hard conversation does: describe who you're asking and what you're working on to an AI, and work through what you expect to hear back, before you're actually in the room (Leadership Explored, episode 8). It won't make the ask for you, but it can get you past your own hesitation before you have to act on it.
None of this replaces the actual ask, or a real second opinion from a colleague or a coach.
7. This Is a Fit Question, Not a Recipe
The shape above is a strong default. Where you personally sit against it, and what that's costing, is what the Engineering Ways of Working Diagnostic is built to find.