The same decision can sit at two different levels of delegation depending on nothing more than what breaks if it’s wrong. Same person, same team, same amount of trust in them. Change what a wrong answer costs and the right level moves.
Most delegation advice sorts the levels by the person instead. Junior people get told, senior people get trusted, and you climb the ladder as someone earns it. Sort them by the failure mode and the choice gets easier, mostly because you can actually name a failure mode.
If you searched for the five-level model, that’s a different post: Levels of Delegation. This is Jurgen Appelo’s seven-level model, and I’m giving you the actual words for each one.
Appelo built the seven levels into Management 3.0’s Delegation Poker, where teams play them as cards: Tell, Sell, Consult, Agree, Advise, Inquire, Delegate. Humanizing Work has a long walkthrough of the model if you want the theory in depth. What’s below is the working version. I work across sales, marketing, engineering and HR, so these are the sentences I use, and the failure mode that puts each decision where it sits.
The level is not about the person
The right level is a property of the failure mode and the cost of being wrong. The person doing the work barely enters into it.
An illustrative pair, made up to keep it clean. “Change the deploy process” during a live incident belongs at Tell: one person calls it, everyone else executes, and the cost of a debate is measured in minutes of exposure. “Change the deploy process” as a quarterly improvement belongs at Consult, because the failure mode there is a technically good change that nobody adopts, and you only find that out by asking first. Same manager, same engineers, same decision on paper.
That’s also why delegation is worth treating as a skill with a measurable shape rather than a personality trait you either have or don’t.
One thing to settle before the levels are useful: are you handing over a task or a decision? If that’s still fuzzy, the art of delegation covers it. The seven levels are all about decisions.
The 7 levels of delegation, and what I actually say
Every scenario below is illustrative. None of them is a real client, a real incident or a real person. The scripts are the part that isn’t invented.
Level 1, Tell
Illustrative scenario: someone flags mid-afternoon that a deploy has exposed a set of customer records. Nobody is asking what the team thinks about rollback strategy. The cost of being wrong is a disclosure notice and a week of phone calls, and every minute spent building agreement is a minute those records stay exposed.
“We’re rolling back to the last stable build in the next ten minutes. I need you on the database migration now. Questions after, not during.”
Seniority has nothing to do with why that sentence is a Tell. A wrong call here can’t be recovered inside the time available, and that is the only thing setting the level.
Level 2, Sell
“This one is decided, so I’m not asking whether we do it. What I owe you is the reason, because if the reason doesn’t land you’ll be back on the old process inside a month.”
Sell is the hardest level to hold, because you’ve given up the speed of Tell without buying the commitment that Consult would have bought you. Illustrative case: a marketing team gets a reporting change it didn’t ask for. The decision is made and it’s mine. What’s still open is whether anybody actually runs the new report in six weeks.
The managers I coach who run cross-functional work live here whether they like it or not. They can’t Tell, because nothing backs a Tell across a reporting line they don’t own, so every call has to be won on the argument alone. That’s a distinct skill, and influence without authority is the version of it worth practising. It also explains why cross-functional work feels slow: the level got fixed by the org chart before anyone looked at the stakes.
Level 3, Consult
The trap at Consult is asking after you’ve already decided. That’s a Tell wearing a Consult costume, and people work it out faster than anyone expects.
Some version of this comes up constantly in coaching: “I asked the team and they were fine with it.” Push on it and the asking usually happened after the decision, in a meeting where the manager had already said what they were going to do. They’re not being cynical about it. They genuinely wanted the input, and they wanted it at the exact point where they could no longer use it.
The illustrative case here is a sales one, a bonus structure that’s producing the wrong behaviour.
“I’m looking at changing the SDR bonus structure next quarter. Before I decide anything, tell me what breaks about the current one from where you sit.”
The giveaway is the second sentence. “Before I decide anything” is a commitment that the decision is genuinely still open. If it isn’t, don’t say it.
The half of Consult that gets skipped is what happens afterwards: going back to the people you asked and telling them what you did with their input, including the parts you set aside and why. Miss that and you get one round of honest answers, then a lot of polite ones.
Level 4, Agree
Agree is the only level where you and the other person land on one answer together and both own it. That makes it the level people claim most often and reach least.
The honesty test: can you overrule the outcome? If you can, and in a bad month you would, it was never Agree. It was a Consult with better manners.
Illustrative, from HR: a performance calibration where a manager and I have to land on ratings neither of us will walk back later.
“Let’s go through the ratings together and land on a number for each person we can both defend. If you disagree with where we land, we’re not done yet.”
That second sentence is what makes it Agree, and it’s the one people leave out, because it’s expensive. It means the meeting doesn’t end on schedule.
Level 5, Advise
“You decide. Here’s my opinion and here’s where I’ve seen that choice go wrong, and I’ll back your call either way.”
Three sentences, and one of them is a promise. Advise also gets picked for the wrong reasons more than any other level on the list, and I’m coming back to that at the end.
Level 6, Inquire
The difference between Advise and Inquire is when you speak. At Advise the opinion arrives before the decision. At Inquire you stay quiet until it’s done.
What puts a decision here is a failure mode you can absorb once but not repeatedly. The thing worth protecting is the quality of the next ten.
“You don’t need me for this one. Once it’s done, I want ten minutes on what you weighed and what you ruled out.”
Those ten minutes are where you find out whether a good result came from good thinking or from luck, and whether a bad result came from bad thinking or from a reasonable bet that didn’t pay. That distinction turns out to be harder than the model admits, and that’s the next section.
Level 7, Delegate
Two things stay out of the handover at every level, including this one: the values and the boundaries. You say them once, and then you stay out of the way.
“This is yours. I don’t need to hear about it unless something’s on fire.”
Delegate is the right level when a wrong answer is affordable and reversible, which covers a much narrower set of decisions than the word suggests. The phrase “unless something’s on fire” also has to mean something specific, agreed before you walk away, or it means nothing at all.
Most managers who believe they’re at Delegate are at Inquire, because they’ve kept a standing check-in. A standing check-in is Inquire. That’s a perfectly good place to be. Just call it by its name, so the other person knows which one they’re in.
What the model doesn’t tell you
The seven levels assume a decision holds one level from beginning to end. It doesn’t. The level moves when the stakes move. A project you handed over at Delegate hits a question about a customer contract, and it’s a Tell by morning, not because you lost confidence in anyone but because the failure mode changed underneath you. The model doesn’t warn you this is coming, so when it happens it reads to the other person as a manager taking something back.
It runs the other way too. Delegating higher than the situation warrants reads, from the other side, as an absence. Kalan Norris and James Lemoine at the University at Buffalo School of Management found that when delegation reads to employees as laissez-faire leadership, it provokes what they call dysfunctional resistance, and in 2021 research on why employees resist delegation they put the cost of that resistance to U.S. companies at about $50 billion a year. Half-hearted effort. A request somebody forgets.
The harder gap is the one Level 6 opens. You delegated at Inquire, and the decision came back wrong. Now what?
In those ten minutes you’re trying to separate two things that produce identical results: sound reasoning that met bad luck, and reasoning that was the problem. They need opposite responses. Pull the level back after one unlucky call and you’ve taught the person that outcomes are what you’re grading, which makes the next three decisions worse. Leave it where it is through a genuine run of bad thinking and you’ve stopped managing the decision at all. I don’t have a clean rule for telling them apart in the moment, and neither does the model. The people explaining the decision to you have an interest in the first story, and they usually believe it.
Why Advise is the level everyone picks
Read the seven scripts again and you’ll notice they’re not sorted by how much I trust the person. They’re sorted by who carries the cost when the call is wrong. Advise is the only level on that list that never answers the question, which is exactly why it’s the one managers reach for when they haven’t asked it.
Go back to Level 5 with that in mind. Advise puts the outcome on the other person while the decision still feels like yours. You gave them the call. You also gave them your opinion, on the record, before they made it. If it goes badly, there is a version of events in which you said so. That’s why it reads as respect from the inside and as a hedge from the outside, and why the same conversation can look generous to the manager and evasive to everyone else in it.
I’ve coached over 300 managers and leaders, and the ones who sit at Advise by default are almost never the careless ones. They’re the careful ones. They’ve read enough about autonomy to know that telling people what to do has gone out of fashion, and Advise lets them hand over the decision without ever handing over the argument.
So here’s the question I ask when someone can’t explain why they picked a level: what failure does this level protect against? If there’s no answer, the level picked itself, and left alone it picks Advise almost every time.
Take the hardest call on your desk this week, the one you keep not making. Write down what happens if the answer turns out to be wrong: who finds out, how quickly, and how much of it can be undone. Pick the level from that, then use the script for it word for word the first few times, because the words are where the level actually lives.
Say which level you’re at out loud before the conversation starts, not after it goes sideways. And if you want a read on where your habits sit before you try it, the delegation assessment takes a few minutes and tells you where yours is breaking down.