against socratic management

use top-down feedback to train the eye (not the hand)

I’m excited to be speaking at Elevate, monday.com’s flagship conference that brings ambitious leaders and doers to the forefront of AI for work. My panel, called “Managing people in the agentic era,” will explore practical ways to use AI to become a more effective manager, while also examining how to lead organizations where people and Agents work side by side.

Elevate takes place in New York on October 28-29. You can grab tickets here; and use code hilarygridley30 for 30% off.

The time has come. I have a rant deep inside of me that I must let out.

DOWN WITH SOCRATIC MANAGEMENT!

“Socratic management” is what I call the school of management that leans heavily into coaching direct reports through open-ended questions.

You go to your manager because you are stuck. “What do you think I should do here?” you ask. They respond, “Well, what do you think?”

This has always vexed me. Presumably, my manager has learned a thing or two over the years, and I would love if they would share some of that wisdom with me.

Then I became a manager and took a variety of manager trainings. Imagine my surprise when I learned that this was not just an irritating habit. It was actively taught as the best way to handle questions!

The reasoning, as far as I can tell, goes something like this:

1. Managers need their teams to get better at solving problems on their own.

2. If you step in with an answer every time someone needs help, they will not develop that skill. They will just learn to bring every problem to you. This will create more work for you and prevent your team from growing.

3. So when someone asks for help, you should resist giving them the answer. Instead, you should “coach” them toward finding it themselves, usually by asking open-ended questions.

The answer, presumably, is somewhere deep in their heart, waiting to emerge.

This presents a false binary: either you solve the problem for them, or you hoard your own thinking like precious gold that nobody can touch or even look upon. Surely there are some options between those two things. There’s gotta be a better way / secret third thing etc.

I really did give this a try. I laugh now thinking about some of the first people I managed, who had to endure a series of increasingly strained questions from an increasingly agitated Hilary. I obviously had an opinion, but instead of sharing it, I had invented a game of “guess what I am thinking,” and for what?

A few times I finally broke and asked, “Is it annoying that I am asking you open-ended questions when I obviously have an opinion?” They laughed awkwardly.

…My management skills have come a long way.

Anyway. The management trainings are right about one thing: you do not want your team to rely on you to solve every problem, or to repeatedly defer to your judgment instead of developing their own. But open-ended questions are not the only alternative to handing someone the answer.

THERE IS, IN FACT, A BETTER WAY.

There are two ways to provide feedback. One is bottom-up: you start from the work or situation at hand, and discuss how to improve it. The other is top-down: you start from a good example of the work, explain the ways in which it is good, and then talk about why the work at hand does not meet that bar for good.

(I mean “top-down” in the literal sense: you start with the standard and work your way down to the specific work. I do not mean command-and-control management.)

Most managers, if they are giving feedback at all, are giving bottom-up feedback. It is reactive: something lands on your desk, and you respond. Change this! Fix that! Love it! LGTM! ¹

The problem is that you have to do this every time. When you orient your feedback around fixing one specific piece of work or solving one specific problem, it is not easily generalizable. It provides no insight into how to solve a similar but different problem. If the job of a manager is to create leverage, this does the opposite: it creates a bottleneck. So I understand why management trainings discourage this behavior.

But. But! The solution is not, go girl give us nothing. Open-ended questions may solve the micro problem of “managers absorbing their team’s cognitive labor,” but they do not solve the macro problem of “a team with an uncalibrated bar for excellence.” They provide no clarity around what good looks like.

That is what top-down feedback is for. You collect examples of good work, share them with your team, and explain exactly what makes them good. There is a reason people love fashion influencers who break down outfits for their audience, who say, notice the proportion here, notice the color harmony. They are teaching you what to look for, which trains your eye.

Let’s say you manage a team of video producers. Someone brings you a video they are working on and asks what you think. You could say: “Here is what I think a good video needs to do. It needs to stop a person who is scrolling and capture their attention in the first two seconds. It needs to identify a pain that the audience acutely feels. And it needs to make its point with nothing feeling slow.” Then show them a video that does those things well and voice over what you are observing.

Then talk about their video. Maybe taken as a whole it is engaging, but the opening does not meet the bar you described. You can tell them that, without telling them exactly how to fix it. If you both are on the same page about what they are aiming for, they can figure it out.

The place where you collect these examples does not need to be fancy. I usually keep a running Google Slides deck where I drop in screenshots of work I come across that I like, along with a few notes about what makes each example good. Once I have enough, I might organize them into something more thoughtful, like this guide to leveling up PowerPoint presentations that I made a few years ago.

It takes time to build up the collection, but very little time on any given day. For the next two months, whenever you come across good work, take a screenshot and drop it into the file. By the end, you will be in great shape.

I did not invent this approach. It is how design critiques work: you break the work into facets, talk about what good looks like for each one, and then evaluate the work against that bar. Design school runs on this kind of feedback.

So design managers should be extraordinary at giving feedback.

And yet design managers rate as some of the most ineffective. Why?

Developing your own taste is only half of the equation when you’re working inside a company. The other half is whether you can articulate what good means for the business you work inside.

A design manager can have a strong point of view about what makes work good, and still fail their team if the team repeatedly meets that bar only to watch the work get killed. It is very stressful for teams to feel like they have to triangulate between multiple different definitions of good, like it’s on them to resolve conflicts between their manager and others in the organization. Mom and dad are fighting! etc.

This is not to say, abandon your principles, optimize for whatever the boss wants. If you are in a leadership position, you are being paid to bring your standards into the organization. But you also have to know what the business needs, where your standards are in tension with that, and what to do about that tension.

This is, of course, hard. Figuring out what good looks like is hard enough; figuring out what good looks like here, given the specific context of the market and various personalities and idiosyncrasies of a company, is even harder.

That’s management, baby!

Once you have collected examples of good work and explained what makes them good, you have also done most of the hard work required to build an AI tool that gives useful feedback.

In my course, we spend a chunk of time building tools like this. People often assume the first step is to connect AI to all their “stuff” — their Drive, email, Slack — and say, “Learn what I think is good, then go evaluate things for me.”

I don’t know what your Google Drive looks like, but diving into mine would not send a particularly helpful signal about my taste. An intern digging through it would not emerge with good judgment, but instead, fear and confusion. It contains polished work, bad work, abandoned work, and things I forgot existed. It has a pitch deck I made to convince my friend to visit Vancouver with me where one of the killer arguments is “Joshua Jackson is from here <3 Pacey <3”. We never made it to Vancouver.

My Google Drive, ultimately, does not represent my point of view. My curated collection of good work does. So, I can give that collection to the robot, and ask it to help me find the patterns. What are my implicit standards, given what I like? This can be very helpful if you struggle to put words around something, yet feel it in your bones.

You can ask the robots to help you distill and articulate clear criteria for excellence. You are, essentially, making an evaluation rubric. Once that is clear, then you can build a tool to evaluate a specific type of work at scale.

You cannot outsource the development of a point of view. If you do not know what good looks like, the AI will not save you.

Plus, you don’t even have to build a tool! In doing this exercise, you will emerge with a clearer point of view and a much better way to explain it to your team. That alone will make you a better manager.

One thing I love about this style of management is that it can be very collaborative. You do not have to be the sole contributor to the shared definition of what good looks like. Everyone can add examples of good work to the file, and you can have conversations about what you collectively think works. These conversations will sharpen your own point of view, as well.

Plus, chances are you have people on your team who are stronger in places you are relatively weak. Maybe you have a strong aesthetic sense, but shaky technical instincts. Find someone who has the inverse. Ask them to show you examples of good API design, or technical documentation, or system architecture, or whatever. Ask them to explain what makes them good. See if you can get to the level of specific facets, or criteria. Do this with your team, and you will improve not only your own technical judgment, but theirs, too.

As a manager, you do not have to be the world’s foremost expert in every part of the work. Lord knows I am not. Your job is to make sure the team, collectively, has a shared bar for good work, and is raising that bar over time.

This is also how you develop future managers. Once someone can move beyond producing good work to explaining why the work is good, they can start teaching that judgment to other people.

If you have a manager who responds to every request for feedback by asking what you think, try changing what you ask for.

Instead of, “Can you give me feedback on this?” ask, “Do you have an example of what you think really good work looks like here? Can you tell me what is good about it?”

This is, yes, an annoying amount of managing up. But most managers do have some kind of standard in their head, even if it is incomplete and they have never put it into words. Asking for an example is often the fastest way to get it out of them.

When you give feedback, you can train someone’s hand, or you can train their eye. Training the hand helps them fix this one slide, video, or plan. Training the eye helps them see what needs fixing the next time. This is how they become less dependent on you — not by learning to answer a series of increasingly vague questions, but by developing judgment of their own.

Train the eye. The hand will follow.

xoxo,
hils

P.S. I’m leading a workshop for parents who want to learn how to use AI to better manage their household. Enroll here.

---

¹ I don’t know if anyone else will find this funny, but I had been using “LGTM” (looks good to me) for about a year before one of my treasured Design colleagues told me she had assumed it meant “let’s go to market!” and was some weird girlboss speak I had invented. I don’t say that anymore.