Case: Klasna's lead segment for 2027

Start where the favorite starts, then put it in its place on the sheet. Size is a real criterion; it's one of six.

From the year by segment and month (September 2025 – August 2026), sum each segment's contribution: exam_prep 7,173,637.12 UAH; younger 2,220,753.08; adults 1,353,985.33; diaspora 1,663,799.19. Total 12,412,174.72. Exam prep's share: 7,173,637.12 ÷ 12,412,174.72 = 57.8%.

On size alone Uliana is right: exam prep gets 5 points on contribution (≥ 5 M). But contribution carries 25% of the weight. The other 75% asks how seasonal the work is, how many families stay, how many enquiries Klasna wins, what each new family costs and how contested the segment is.

The sheet was fixed before anyone looked at a segment: criteria, weights, directions and bands. That's what lets the size number count fully without deciding the result on its own.

Peak share measures how much of a segment's year falls into a few months: here, lessons in February–May as a share of the year's lessons. Lower is better, because steady work is cheaper to staff.

Exam prep delivered 22,176 of its 40,855 lessons in February–May: 22,176 ÷ 40,855 = 54.28%. On the sheet's bands (> 50% = 1 point) that's the lowest score. Half the year's work in four months means extra coordinators and tutors in spring and idle tutors in July.

The families in the EU sit at the other end: a peak share of 34.54%, 5 points (≤ 36%). Their lessons are spread across the year, which is part of why they're cheap to serve.

Seasonality is invisible in annual totals: two segments with the same contribution can cost very different amounts to staff. That's why it earns its own line on the sheet.

Cost per new family = marketing spend aimed at the segment ÷ new families it brought, over the same period. Lower is better.

Sum the twelve monthly rows for each segment, then divide once:

• diaspora: 265,420 ÷ 94 = 2,823.62 UAH → 4 points (2,000–3,000)
• exam_prep: 1,522,900 ÷ 435 = 3,500.92 → 3 points
• younger: 597,195 ÷ 125 = 4,777.56 → 2 points
• adults: 464,485 ÷ 120 = 3,870.71 → 3 points

Divide the yearly totals rather than averaging twelve monthly ratios: months with very few new families would swing a monthly average wildly.

Read it next to the win rate (new families ÷ enquiries): exam prep converts well (435 of 1,207 enquiries, 36%, 4 points) because teachers send families already convinced, yet each family still costs more than a diaspora family.

With the sheet fixed, scoring is mechanical: turn each value into points with the bands (a value exactly on a bound takes the better score), multiply by the weight, add up.

Diaspora:

0.50 + 0.75 + 0.75 + 0.45 + 0.40 + 0.80 = 3.65.

The others: younger 3.40, adults 3.05, exam prep 3.00 (5, 1, 2, 4, 3, 2). The biggest segment ranks last: it wins on size and win rate and loses on seasonality, retention and competition. Small, steady, loyal and cheap to win puts the families in the EU first.

Uliana's draft used the same six criteria, but scored "contested" the wrong way: more rivals, more points ("proven demand"). Reversing a 1–5 score means 6 minus the correct points, so 2 becomes 4, 4 becomes 2, 5 becomes 1. With a 20% weight, each two-point swing moves a score by 0.40.

• exam_prep: contested 2 → 4, so 3.00 → 3.40
• diaspora: 4 → 2, so 3.65 → 3.25
• younger: 4 → 2, so 3.40 → 3.00
• adults: 5 → 1, so 3.05 → 2.25

Exam prep jumps from last to first: one arrow the wrong way puts the most contested segment at the top. Demand was already counted under contribution, so the draft counted it twice and erased how hard each exam family is to win (59.38% named a paid alternative).

Choosing a lead segment decides where new effort goes: campaigns, offers, the words on the main page. It doesn't stop serving anyone. The plan row should say, segment by segment, how each is still served, so nobody reads the choice as abandonment.

For Klasna in 2027: diaspora families lead. Exam-prep families, 57.8% of contribution, keep coming the way most already do, through school teachers (42.19% of exam accounts first heard from one), and get exactly the same lessons. Younger learners' families, close behind on the sheet at 3.40, are kept warm through referrals. Adults keep finding Klasna through search.

That's the answer Leo needs in his own terms: the biggest segment doesn't shrink because it isn't the lead; Klasna simply stops adding marketing money to the most contested, most seasonal group and puts it where each hryvnia wins more.

A scoring sheet supports a choice; it doesn't prove it. The memo should state how sure you are in plain words, next to the choice, not in a footnote.

For Klasna's lead segment that means four things:

• The margin: diaspora 3.65, younger families 3.40. Close enough that a different but defensible weight could change the order.
• Small samples: two criteria (still active, contested) come from the customer export, which holds only 26 EU accounts.
• What the sheet leaves out: diaspora families pay in euros, a currency risk no criterion measures.
• When it will be checked: a review date for the choice.

Extra decimals don't add confidence; honest limits do. They also protect the plan: if the June numbers move, nobody is surprised.

How it fits together: fix the sheet first; score size, seasonality, retention, win rate, cost and competition; watch the direction of every arrow; lead without leaving; say how sure you are. Next module: once Klasna knows whom it's for, why should they choose it over their alternatives?