From words to money: ladders, pre-orders, deposits

A survey records what people say. A price ladder records what they do: the same offer shown at several real prices, one price per comparable group, counting who actually commits money at each step.

For NMT Sprint, Nazar's team took three comparable groups of 150 parents with maths packages and offered each one price inside the corridor, 13,000, 12,400 or 11,800, with a refundable deposit of 1,000. Every step was a price Klasna would really charge.

Two design choices matter. Each group sees one price, so the test measures response to a price, not a choice between options on one page. And the commitment costs something real, a deposit, so the count reveals willingness to pay instead of stating it.

Showing different groups different prices raises a fair question: what if your group saw a lower price than the one finally chosen, or a higher one? A ladder is honest only with a clear promise, given to everyone up front.

Hanna's promise works both ways:

• if the final price is lower than the one you saw, you pay the final price;
• if it ends up higher, you keep the price you saw or get your deposit back in full.

So a family that left a deposit at 11,800 when the final price is 12,400 keeps 11,800 or takes its 1,000 back. Nobody pays more for having been put in a particular group, and the deposit is refundable in practice, not just on paper.

Before the deposit link, the 12,400 group got one question: would you buy? 48 of 150 parents said «definitely» or «probably». Then 12 paid the deposit.

12 / 48 = 25% of the stated yeses turned into money. Put the other way round: 32% said yes (48 / 150), but 8% paid (12 / 150). The stated answer overstated real commitment four times.

That gap isn't dishonesty; it's how stated answers work. Saying yes costs nothing; a deposit costs 1,000, even refundable. Use stated interest to spot direction, and only real commitments to size demand.

The ladder's results: 13,000: 9 deposits of 150 (6.00%), 12,400: 12 (8.00%), 11,800: 14 (9.33%). On revenue, 11,800 wins: 165,200 against 148,800 and 117,000. Nazar's first reaction was exactly that.

But revenue isn't what Klasna keeps. Each sprint carries 8,174.40 of variable cost, and the fee takes 2.5% of the price. So compare contribution per step, assuming every deposit becomes a sprint:

contribution = deposits × (price × 0.975 − 8,174.40)

• 12,400: 12,400 × 0.975 − 8,174.40 = 3,915.60 per sprint; × 12 = 46,987.20.
• 11,800: 3,330.60 per sprint; × 14 = 46,628.40.

The step with the most buyers isn't automatically the one that earns the most.

On contribution, 13,000 trails (40,505.40), while 12,400 (46,987.20) and 11,800 (46,628.40) are almost tied. Is 12,400 the winner?

Not proven. Twelve deposits against fourteen is two families; one more cancellation and the order flips. With 9–14 deposits per group, the gap between the two lower steps is noise. A ladder this small points a direction: 13,000 looks too high, the lower two are close.

Then use what the numbers can't show. 11,800 sits only about 80 above Hanna's target floor of 11,719.38, so it would take a tutor's scarce spring hours for almost nothing extra. 12,400 earns 585 more per sprint and stays clear of that floor. A low deposit rate months before the start is normal, not a failed test.

Sometimes the offer doesn't exist yet. Leo wonders whether parents would buy a summer programming sprint for teens. A fake-door test answers that cheaply: a page with the real description and the real price, and a button. Count clicks per visitor.

The honest part comes right after the click: say plainly that the sprint isn't open yet and offer a waitlist. Three rules keep it honest:

• never take payment for something that doesn't exist;
• never say «sold out» when nothing was for sale;
• keep the price on the page the one you'd really charge, or the clicks measure interest in a price you won't offer.

A click is a real action, but it costs the visitor little, so it's a weaker signal than a deposit.

Signals of willingness to pay differ in strength, and the rule is simple: the more a signal costs the person, the more it reveals.

From strongest to weakest:

1. a paid purchase at the real price;
2. a refundable deposit at the real price: money moved, even if it can come back;
3. a click on «Join» on a page showing the real price: effort, no money;
4. «definitely» in a survey: costs nothing;
5. an answer generated by an AI model playing a customer, which isn't anyone's willingness to pay at all.

Use the strongest signal you can afford for the decision at hand. For a launch price, deposits or pre-orders; for an early idea, a fake-door click may be enough.

Leo's shortcut, "take real orders and refund them if we don't launch", can be honest, but only on stated terms. An honest pre-order says up front: it's a pre-order, it starts if enough people join, by which date, and with a full refund otherwise. And the refund is as easy as the payment: one message, no forms.

Practices that keep a test honest: the real price; the truth right after the click; money only under clear, refundable pre-order terms. Practices that turn it into a trap: a fake «sold out», or a deposit charged without saying it's refundable.

How it fits together: stated yeses overstate commitment, so price ladders and deposits measure what people pay; judge steps on contribution, read small samples as direction, and keep every test honest. The case that closes this module returns to Club Plus: the survey, the people it never asked, and a real deposit.