Levels of freedom: matching person and task

How much freedom to give isn't a property of a person; it belongs to a person on a task. Someone can be an expert on schedule changes and a beginner on refunds, and should get different freedom on each.

This is the dial behind autonomy at work: how much say a person has over how, when and in what order they work. More autonomy tends to raise motivation and learning, but only with clear goals and the skill to use it.

Get the level wrong and it hurts either way. Too low, and an expert waits for approvals, stops thinking and slowly disengages; the lead becomes a bottleneck of sign-offs. Too high, and a beginner guesses, and families pay for the guesses.

So the instinct “Taras closes the most, give him the most freedom; Dmytro and Sofiia are new, give them the least” is half right at best. Volume on one type of ticket says little about judgment on another, and newcomers bring skills from previous jobs.

A simple scale makes freedom explicit, so nobody has to guess:

1. Gather the facts, I decide.
2. Propose, I approve.
3. Decide, tell me after.
4. It's yours; we review it monthly.

The level follows experience on that task and the cost of a mistake. At Klasna: Dmytro, on day 3, facing a family's refund request: level 1. He knows neither the policy nor the system yet, and money leaves on a refund. Maryna, who knows refunds but hands most of them up: level 2, a stretch. Taras on schedule changes, 900 solved with 4% reopened: level 3. Olha on refund exceptions within the policy: level 4, agreed in January after a long track record and a conversation, and reviewed rather than approved.

Levels can differ within one person, and they should move. A level is a starting point that rises as experience builds, not a rank.

To set levels fairly, use a rule anyone can check, per person and per ticket type. At Klasna the starting rule: at least 50 solved with under 5% reopened → level 3; at least 20 solved → level 2; fewer → level 1. Level 4 never comes from a table alone; it takes a conversation.

Two measures help read the six-month table. The solo rate, solved / (solved + handed up), shows how often the person finishes a type without passing it up. The reopen rate, reopened / solved, shows how often their answers hold.

Taras on schedule changes: 900 solved, 4.00% reopened, 900 / 903 ≈ 99.67% solo. Level 3, easily. Taras on refund requests: 60 solved, 70 handed up, so 60 / 130 ≈ 46.15% solo, and 15 of his 60 (25.00%) came back. He hands up more refunds than he solves. Level 2: he proposes, Olha approves.

Same person, two levels. And as Iryna reminds you, refunds are where money leaves the company.

A per-task table often overturns the impressions a lead carries from one headline metric. Fast first replies and deep skill on a ticket type are different things.

At Klasna, Bohdan has been the slowest replier since January, about 35 minutes. In the six-month table he turns out to be the team's video-failure expert: 160 solved, only 2 handed up, so 160 / 162 ≈ 98.77% solo, with 3 reopened, 1.88%. Level 3, and the natural owner of those tickets from 8 March.

The same table holds other surprises. Overall, Taras starts at level 3 on one type of six. Olha starts at level 3 on all six and solves 59.15% of the team's refund requests (210 of 355). Maryna is at level 3 on four types, Bohdan on five.

The lesson for a lead: before deciding who gets freedom, look at the task-level data, not the ranking in your head. The slowest replier may be the person the team most needs on its hardest kind of ticket.

The starting rule only works if you apply it to each person-and-type cell, with both conditions for level 3: at least 50 solved and under 5% reopened.

Worked through at Klasna:

• Bohdan, video failures: 160 solved, 3 reopened, 3 / 160 ≈ 1.88% → level 3.
• Olha, refund requests: 210 solved, 8 reopened, 8 / 210 ≈ 3.81% → level 3.
• Maryna, Teams invoices: 20 solved → level 2 (exactly the threshold; 1 reopened, 5.00%).
• Taras, payment questions: 700 solved, but 63 reopened, 9.00%, not under 5% → level 2, despite the volume.
• Dmytro, any type: no Klasna history, fewer than 20 solved → level 1.

The common slips are skipping the reopen condition because the volume is impressive, and giving a newcomer a middle level because level 1 feels unwelcoming.

The table is honest, but it's a starting point. The next step adds what a table can't know.

A data rule is a fair start, not a verdict. Two more inputs decide the final level.

The cost of a mistake. A wrong schedule change costs a rebooking; a wrong refund costs money and a family's trust. Where mistakes are expensive, step up more slowly than the table alone suggests.

The person's own view. Ask: “On refunds, how much support do you want right now?” Some people want more freedom than their numbers show and will grow into it; others are skilled but not yet confident and want a safety net for a while.

Then say each level out loud, and in writing. In a remote team nobody watches over a shoulder, and nobody should. The level written in the brief is what replaces guessing: the person knows where they decide alone and where they propose, and so does everyone who works with them.

At Klasna that means a short table in each brief: ticket type, level, and what moves it up.

Newcomers start at level 1 on most things, but a level is a starting point, not a season-long sentence. Plan the climb from day one.

At Klasna, Dmytro and Sofiia joined on 1 March with no Klasna history. Level 1 at first is right. But Sofiia did a season of support at another online school, so on payment questions she may climb fast. A sensible plan: level 2 on routine types (schedule changes, payment questions) after week 2, if the reopens stay low, with money decisions like refunds stepping up later.

Keeping both at level 1 on every type for the whole spring peak may feel safest, but it isn't. It routes every one of their decisions through someone else for three months, which overloads the experienced people in the busiest weeks of the year. And it stops the newcomers learning, so they arrive at May as dependent as in March.

Write the climb into each newcomer's brief: which types, which conditions, which date to look again.

The same level can feel like a demotion or a stretch, depending on how it's set. The difference is whether the person wants it and knows what the next step takes.

At Klasna, Maryna on refunds: 40 solved, 45 handed up, 5.00% reopened. Level 2 by the rule. Olha adds what the table can't: Maryna is good at refunds but doesn't trust herself yet; Olha offers to approve her proposals for a month.

So level 2 on refunds is Maryna's stretch. The best next step: ask Maryna whether she wants it, then write level 2 and the step up to level 3 into her brief: what level 3 would take, and when you'll look again. Assigning it unasked skips her view; blocking refunds stops her practice; jumping to level 3 on Olha's word ignores the data and the cost of a refund mistake.

How it fits together: freedom belongs to a person on a task; four levels make it explicit; a data rule sets a fair start; cost, the person's view and a planned climb finish it. The next lesson designs work people grow in, not just work that gets done.