Segment prices with fair fences

Charging different groups different prices is called segmented pricing, and it's everywhere: student prices, sibling prices, early-booking prices, off-peak prices. It's normal and lawful in general, and it lets a seller serve people who would never pay the full price without cutting it for everyone.

What makes it fair is the price fence: the rule that decides who pays what. The problems lie elsewhere: prices personalised in secret, collusion between competitors, abuse of a dominant position and unfair practices. In the EU, sellers to consumers must disclose a price personalised by automated decisions.

Leo's idea, charging each family what it can pay based on its device, visits to the price page and where it comes from, fails on the fence, not on the idea of different prices. It's hidden, it can't be explained to a parent, and it wouldn't stay hidden: parents compare prices with each other, and the first screenshot ends it.

So the question for every segment price is: can you state the fence openly and defend it?

A price fence works when it passes three tests:

• Real: it describes a true difference between buyers (a second child, a student, a time of day), not a guess about what they can afford.
• Easy to verify: the check is quick and, ideally, done once.
• Respectful: it asks nothing awkward and humiliates no one.

“10% off a second child's packages in the same family account” passes all three: the second child is real, the account checks it by itself, and nobody has to explain anything.

The failures are instructive:

• a higher price for visitors on expensive phones: undisclosed personalisation, and indefensible once seen;
• a lower price for parents who say the price is too high: it rewards complaining and can't be checked;
• a secret lower price for one district: not transparent and hard to defend when another family asks.

A fence you'd be uncomfortable explaining to a parent is the wrong fence.

Every fence needs a check, and the check has to be proportionate: quick, done once where possible, and asking for no more information than the condition itself. It also shouldn't cost more than the discount.

Klasna's fences and their checks:

• Sibling price: both children are in the same family account. The account does it.
• Student price for adult learners: a valid student card, checked once a year, not with every package.
• Regional price: where the family lives and pays, which the payment already shows.
• Prepayment price: the whole package is paid up front.
• Social price for displaced and veteran families: one document check, no explanations required.

Regional prices are lawful in general, but inside the EU be careful with different terms by nationality or residence: before giving two EU countries different prices, check the rules for that case with a lawyer. That's a dated guardrail, not legal advice.

A social discount is different from the trades of the last lesson: it doesn't buy anything. It's a choice Klasna makes and pays for, because it wants displaced and veteran families to be able to study. That's a fine reason, as long as Klasna knows the cost.

HELP20 gives those families 20% off. On the cheapest package per lesson, the English 32-pack at 580, the price becomes 464. After the 2.5% fee, 452.40; minus 420 for the tutor, video and AI, 32.40 UAH a lesson is left, just above the 430.77 floor price. So HELP20 is affordable on its own, with little room for anything on top.

The fence must respect people:

• one document check, once, not with every package;
• no stories and no explanations required;
• no public lists, and nothing kept beyond the word “eligible”: no document copy.

A social discount that humiliates the people it's for fails its own purpose.

Each discount can be affordable on its own and ruinous together. A family can be both displaced and have a second child, so Hanna asks what happens when HELP20 meets SIBLING10.

On the English 32-pack at 580 a lesson, stacked as 30% off: 580 × 0.7 = 406; after the 2.5% fee, 395.85; minus 420 for the tutor, video and AI: −24.15 UAH a lesson, −772.80 on the package. Even multiplied (0.8 × 0.9 = 0.72), the price is 417.60 and the lesson still loses 12.84.

HELP20 alone left 32.40 a lesson. There simply isn't room for a second discount on the cheapest package.

So the rule is: discounts don't stack; the larger one applies. Klasna's codes already work this way. It's easy to explain (“you get the bigger of your discounts”) and it protects every package from going below cost.

Before launching any new discount, check what it does when combined with every existing one on the cheapest package.

A learner offers Dana a deal: the month paid in advance for 5% off. A discount should buy something, so ask what prepayment buys her. Her hope: fewer no-shows.

Price it over her year. She expects 348 billed lessons at 600. If every learner paid monthly ahead at 5% off, the discount would cost 348 × 600 × 5% = 10,440 UAH.

What it would fix: 21 lessons this year where the learner doesn't come and she doesn't charge, worth 21 × 600 = 12,600 at full price. But a discount doesn't recover those lessons; a rule does. “Lessons cancelled less than 24 hours ahead are paid”, announced before booking, gets that money back without giving 10,440 away.

Prepayment can still earn something that costs her nothing: a first pick of evening slots. For a learner who wants a fixed weekday evening, that's worth more than 30 hryvnias a lesson.

When a discount is meant to fix a behaviour, check whether a clear rule fixes it more cheaply.

Time of day is a good fence: it's real, anyone can see it, and anyone can choose it. Dana's weekend mornings are always empty, while everyone wants weekday evenings. A lower morning price can fill those hours.

The timing matters. While her evenings still have free slots, a cheaper morning would mostly move learners off 600 evenings onto cheaper mornings: the same lessons for less money. Once the evenings are full, a morning price fills an hour that otherwise earns nothing and doesn't push out a 600 evening.

The numbers after her 6% tax: an evening lesson keeps 600 × 0.94 = 564, a morning one at 540 keeps 507.60. Less per lesson, but more than an empty hour.

And the price is public, on the same terms for anyone who books a weekend morning. That's the difference between a fence and a favour. Until her evenings fill, Dana has only ever charged 600; the 540 is her rule for when they do.

A one-person practice can't negotiate every price in a chat. Dana needs a few public rules she can send to anyone who asks:

• One public price, 600, for everyone, friends included. A secret “friend price” doesn't stay secret, and then every learner wonders what the real price is. If she wants to give, she gives openly and rarely: a lesson as a gift.
• A cancellation rule instead of a prepayment discount: lessons cancelled less than 24 hours ahead are paid, stated before booking; paying a month ahead earns a first pick of evening slots.
• A quiet-hours price anyone can choose: 540 for weekend mornings, opened once her evenings are full.

The failures share one trait: they're favours, not fences. A price for people she knows or for those who complain most spreads by word of mouth and makes everyone else doubt the price.

How it fits together: segment prices are fair when the fence is real, checkable and respectful and the discounts don't stack. Next: promotions, and the gap between list price and the price Klasna actually gets.