Case: redesign Klasna Club for 2027/28

An average answers “how much in total?”. It doesn't answer “who?”. When a product has a cost per use, the second question decides whether a flat price is safe.

Klasna's November export covers all 880 Club members: 44,000 AI minutes, 50 per member, 40 UAH each on average, exactly what the budget assumed. Now sort the members by minutes:

• half use 16 minutes or fewer (the median);
• nine in ten use 81 or fewer;
• the mean of 50 sits far above the median, a sign that a small group uses a lot.

When the mean is much higher than the median, the distribution is skewed: most people are light users and a few are very heavy. For a flat price with unlimited use, those few are the risk, and the average hides them completely.

So the case starts with the heaviest tenth, 88 members, and asks how much of the total they use.

To see whether a few users carry the cost, measure concentration: the share of the total used by the heaviest group. Sort members from most to fewest minutes, take the top slice, and divide its minutes by all minutes.

In Klasna's November data the heaviest tenth is 88 members. Together they used 28,432 of the 44,000 AI minutes: 28,432 ÷ 44,000 = 64.62%. A tenth of the members used almost two-thirds of the AI.

Their AI cost averages 28,432 × 0.80 ÷ 88 = 258.47 UAH a member, more than six times the Club-wide 40. Under a flat price with unlimited use, this group is where the risk sits; the other 792 members cost about 15.73 each.

Concentration like this is normal for products people use by choice, not a sign of abuse: enthusiasts practise a lot because they love it. It just means the price has to be designed for the heavy few as well as for the typical member.

Concentration shows where the risk is; contribution per member shows the actual loss. Work it for every member:

price per month × 0.975 − 55 − 0.80 × minutes

with 450 for monthly members and 360 for annual ones (4,320 ÷ 12). A monthly member turns negative above 479.69 minutes, an annual one above 370.

In November 24 members were below zero: 14 on the monthly plan and 10 on the annual plan. Together they took −6,699.50 UAH of contribution a month. The Club as a whole still leaves 293,725 a month, so the problem is real but concentrated, not general.

Leo calls them a success story, and Hanna answers that it's both: they love the partner, and each of them costs Klasna money every month. Both views are right, which is why the fix should keep these members practising and change only what they pay for the extra, not push them out.

A fair-use limit is an included amount set so that typical users never notice it and the heaviest stop being a loss. Size it from the usage spread, not from the average.

Nine in ten Club members use 81 minutes or fewer, and the median is 16. A limit of 300 minutes a month, about five hours, sits far above all of them and below both break-even points (479.69 monthly, 370 annual).

In November only 31 members (3.52%) went above 300, by 11,674 minutes in total. Capping them would save 11,674 × 0.80 = 9,339.20 UAH a month. At the limit nobody is negative any more: a monthly member leaves at least 383.75 − 240 = 143.75, an annual one at least 296 − 240 = 56.

The limit covers the Club and Club Plus; Club Lite keeps its own 120 minutes. Some heavy users will buy a top-up instead of stopping, which also covers their cost.

A limit needs an answer to “and above it?”. Cutting people off mid-practice feels harsh, and these are the members who love the product most. A top-up, more minutes at a stated price, keeps them practising and makes heavy use pay for itself.

Price the top-up from its cost per use, with a margin. Klasna's draft: 200 more minutes for 190 UAH.

• after the 2.5% payment fee: 190 × 0.975 = 185.25;
• AI for 200 minutes: 200 × 0.80 = 160;
• contribution: 25.25 UAH per fully used top-up.

It isn't meant to be a profit centre. It turns the heaviest users from a loss into a small plus, and it gives them a choice: buy more now or wait for next month's minutes.

Two checks for any top-up: it must cover its cost even if the exchange rate moves a little (the AI is billed in dollars), and its price must be on the price page next to the limit, so nobody meets it for the first time at minute 300.

An annual plan at 4,320 a year is 360 a month against 450 on the monthly plan, which looks like a 20% giveaway. But a month is the wrong unit of comparison: the plans differ in how long people pay.

The Club launched in summer 2026, and from the cancellations so far Hanna estimates that a monthly member stays about seven months. An annual member pays for twelve, up front. So compare what one member leaves over their stay:

• monthly: 349.33 a month on average × 7 = 2,445.31;
• annual: 212.46 a month on average × 12 = 2,549.52.

Per member, the annual plan wins, if narrowly, even with the discount. Prepayment also brings cash in early and removes eleven chances to cancel.

Two cautions. Seven months is an estimate from a young product; recheck it as cancellations accumulate. And notice that annual members leave much less per month than the price gap explains. The next card shows why.

A price discount isn't the only thing that changes when a plan is cheaper per month: the plan can also attract different buyers. A prepaid year of unlimited AI is most attractive to the people who expect to use it most.

That's what November shows. Annual members practise 104.43 AI minutes a month on average, monthly members 43.02. So annual members leave only 212.46 a month of contribution against 349.33, far below what the 90-UAH price gap alone would explain.

The limit changes this. With 300 minutes included, annual members rise to 251.91 a month, or 3,022.92 a year, against 2,493.75 for a monthly member's seven months. The per-member lead of the annual plan grows from narrow to clear.

So the answer is to keep the annual plan at 4,320, with the fair-use limit, rather than drop it (the per-month view) or raise it to 5,400 (which removes any reason to prepay). Unlimited annual use is exactly where the heaviest users gather.

A fair-use limit takes something away from the few who use the most, here 31 members. Say so plainly rather than dressing it up as a gift.

Honest rollout of a change in terms:

• On the price page before anyone pays: the limit, what happens at it and the top-up price, in plain words, never a vague «reasonable use».
• Announced early: Klasna tells current members with the 2027/28 price list in June, not in a footnote in August.
• Prepaid deals honoured: the limit applies from September 2027 to new purchases and renewals; annual members keep their current terms until their plan renews.
• No hidden versions: no silent slowdown or cut-off.

How it fits together: the usage spread found the 24 members who cost more than they pay; a 300-minute limit with a top-up fixes that while 9 in 10 never notice; per-member comparison keeps the annual plan. Next module: showing these prices honestly.