Case: splitting the spring peak

A plan for a busy season should start with one comparison: expected work against capacity, week by week. Everything else, who owns what, what the lead does, what the tools do, is about closing whatever gap that comparison shows.

At Klasna the forecast covers 13 weeks, 1 March to 30 May: 5,140 expected tickets, of which 2,465 are routine (schedule changes and payment questions) and 325 hard. Capacity is the six coordinators' planned tickets a week: the core four at their January pace (Olha 75, Taras 95, Maryna 60, Bohdan 55 = 285), plus Dmytro and Sofiia ramping up from 25 to 55 each. Once the new two are at full pace, capacity stays at 395.

Starting from the gap avoids two common mistakes: assigning owners before you know how much work there is, and assuming “everyone works a bit harder” will cover it. It also shows when the problem arrives, which decides when each fix has to be in place.

With expected tickets and capacity side by side, the gap is simple arithmetic: for every week where expected work exceeds capacity, subtract; then add the short weeks up.

At Klasna capacity covers the forecast up to 12 April. From 19 April the expected tickets pass 395 and keep climbing until the week of 17 May. The short weeks: 19 April 405 − 395 = 10; 26 April 20; 3 May 35; 10 May 50; 17 May 65; 24 May 55. Total: 10 + 20 + 35 + 50 + 65 + 55 = 235 tickets over six weeks, worst in the week of 17 May (460 expected against 395, a gap of 65).

Two things matter beyond the total. The gap opens late: the first eight weeks are fine, which makes it easy to relax in March. And it peaks in one week: a fix that covers an average short week (about 39 tickets) won't cover 17 May. Size the fix for the worst week, and have it running before 19 April.

Under pressure, old reflexes come back. Faced with a gap, a lead who used to be the fastest specialist reaches for the familiar fix: “I'll take the hard ones myself.”

Do the arithmetic before agreeing with yourself. At Klasna the week of 17 May has 33 hard tickets. At about an hour each, that's 33 hours: 33 / 40 = 82.50% of the lead's week. Seven hours would remain for one-to-ones, team questions, the weekly look at AI drafts and hand-ups outside policy, in the busiest week of the year. Over all 13 weeks the hard tickets add up to 325 hours.

And the cost isn't only the lead's time. Olha, Bohdan and Maryna would lose exactly the work they're best at, or growing into, just as Your output is now the team's output showed in January: the team stalls and the lead becomes the bottleneck. Olha says it plainly: she saw how that went in January and would rather not watch it again in May.

Being the best at hard tickets is a reason to coach, not to take them.

A tool's contribution to a capacity plan should be computed from stated assumptions, not hoped for.

Two assumptions at Klasna. From the pilot so far, checking an AI draft takes about half the time of writing a routine reply, so each routine ticket handled with a draft frees half a ticket of capacity (rounded down per week). And the lead's 8 own-ticket hours a week, the target week from January, cover about 17 tickets.

The week of 17 May has 223 routine tickets. With drafts: 223 × 0.5 = 111.5, rounded down to 111 tickets of capacity freed. The gap that week is 65, so drafts cover it. The lead's 17 alone wouldn't: they cover only the first short week (a gap of 10). Across the six short weeks, drafts free between 95 and 111 tickets each week, enough for every gap.

The condition matters as much as the number: drafts close the gap only if a person reads, edits and sends every reply. Write that into the assistant's brief and check it weekly.

A gap can be closed in ways that quietly undo months of work. Test each fix against what the team has built: the briefs, the levels of freedom, the one-to-ones.

At Klasna two fixes pass. AI drafts on the 223 routine tickets, each checked and sent by a coordinator, free about 111 tickets, enough for the 65 gap. The lead's own queue time at peak hours on routine tickets adds about 17 as a buffer.

Three fixes fail the test. Letting the assistant send routine replies directly to save the checking time breaks the brief's core limit: a tool can't be accountable, and the first wrong promise lands on a family. The lead taking all 33 hard tickets is January's trap. Cancelling one-to-ones in May removes the support people need most in the busiest weeks, and teaches them that one-to-ones are the first thing to go when it matters.

As Olha notes, Bohdan has started the list of things a draft must never say. The strictest reviewer is part of the fix.

Once the gap is covered, the hard tickets go to people who already own those types, at the level the data supports, the starting rule from Levels of freedom: matching person and task.

At Klasna:

• Refund requests within policy: Olha, level 4, hers, reviewed monthly. She solves 59.15% of the team's refunds and has only just got them back.
• Video failures and Teams invoices: Bohdan, level 3, decides and tells the lead after: 160 video failures at 1.88% reopened, 50 invoices at 2.00%.
• A share of refund cases as a stretch: Maryna, level 2, proposes and Olha approves.
• Schedule changes: Taras, level 3.
• Routine types with drafts: Dmytro and Sofiia, level 2, climbed from level 1 after their first weeks.

Each line matches experience on that type, not seniority or speed. Written down in the plan, the levels also tell everyone where to send what, so nothing defaults back to the lead.

A good peak plan names what the lead does, not only what everyone else does. Otherwise the lead's week fills with whatever falls through the cracks, usually hard tickets.

At Klasna the lead's slots: queue time at peak hours on routine tickets (the 8 own-ticket hours, about 17 tickets a week); hand-ups outside policy, decided with Iryna; the weekly look at AI drafts, reading twenty sent replies each Friday; and the one-to-ones, especially in May, when load and stretch are most likely to tip into overload.

Two things are deliberately absent. The lead doesn't take the video-failure or refund tickets: they have owners. And the lead doesn't approve every refund: Olha decides within policy, Maryna's proposals go to Olha.

Notice how this matches the target week from January: some doing, mostly enabling. The peak is exactly when that balance is hardest to keep and matters most.

When a manager asks “wouldn't it be simpler if you just did it yourself?”, answer with the numbers and with what would be lost.

At Klasna, Iryna asks exactly that about the hard tickets in May. The numbers: 33 hard tickets in the week of 17 May would take 82.50% of the lead's week, 325 hours over the peak. What she'd lose: a lead with no time for the team, and four people who stop growing just when they learn most. Olha adds that she'd like to keep the refunds.

The final page also needs one line per person on how the peak helps them grow, written after asking them, such as Maryna's refund stretch or Dmytro and Sofiia's climb to level 2. A plan that only moves tickets holds for two weeks; one people see themselves in holds through May.

How it fits together: find the gap, size the fix for the worst week, close it without breaking the briefs, give hard work to its owners and name the lead's own slots. In April the peak climbs, and the next module asks what keeps people going when the work gets heavy.