Pricing and Unit Economics — Is There Contribution Margin Left?

In one line

The winner of a pricing experiment is not the price with the highest conversion rate but the price with the largest contribution margin left by each visitor. The value of a channel is not where CAC is cheap but where customers pay back the CAC before they leave. Both are calculated not on revenue but on contribution margin, which is revenue minus variable costs.

Why this was needed

We tested three prices: 9,900, 14,900, and 19,900 won a month. The 9,900 won price had the highest conversion rate and the team settled on it. Three months later customers had grown, but the bank account emptied faster. Payment fees and server and support costs attach in proportion to the number of customers, and the cheap price left little after subtracting those costs.

The same thing happens with ad channels. The customer acquisition cost (CAC) of social ads looked the cheapest. But the customers who came in through that channel left within two months, and the contribution margin they left before leaving barely exceeded the CAC. It was a channel whose losses grew the more you raised the budget. Unit economics reveals this kind of judgment in a single calculation.

How it works

From revenue to contribution margin. In Korea, the price charged to consumers often includes value-added tax. Article 30 of the Value-Added Tax Act sets the tax rate at 10 percent. So of a VAT-inclusive price of 14,900 won, the supply value that remains as the company's revenue is 14,900 × 100/110. Subtract from that the payment fee (this lab sets it as a fixed percentage of the price plus a fixed amount per transaction) and the monthly server and support costs per customer, and you get the monthly contribution margin of one customer. The fee rate and costs are example values from this lab's materials, and real contracts differ from company to company. This module is not tax advice, and the calculation can differ by taxation type (general or simplified).

Judging a pricing experiment. If you showed the three prices to a similar number of visitors, compare the first-month contribution margin per visitor for each price as (purchases − refunds) × monthly contribution margin ÷ visitors. Looking only at conversion rate, the cheap price wins, and looking only at revenue, the price that ignores the cost structure wins. This comparison has limits too — if price also affects the churn rate, the first month alone is not enough, and then you must follow cohorts by price for several more months.

Churn and LTV. The monthly churn rate is "the share of customers who paid last month who did not pay this month". If you assume churn is constant each month, the average time a customer stays is 1/churn months, and so the simplest customer lifetime value (LTV) is monthly contribution margin ÷ monthly churn rate. You must state along with it that this is a simple model that ignores discounting, upsell, and changes in churn. The most common mistake is to put the price (revenue) in the numerator instead of the contribution margin — the LTV inflates by tens of percent, and the ad budget ceiling set with that LTV inflates with it.

CAC and payback period. A channel's CAC is that channel's ad spend total ÷ the new customers who came in through that channel in the same period. A "blended CAC", which divides all ad spend by all new customers (including organic inflow that came in without ads), dilutes the real cost of paid channels with organic inflow. The payback period is CAC ÷ monthly contribution margin, that is, the number of months it takes one customer to repay the CAC. Churn differs by channel, so LTV/CAC must also be calculated per channel.

Write the decision rule first. This lab uses "increase the budget if LTV/CAC ≥ 3 and the payback period ≤ 12 months" as an example rule. The thresholds should vary with the company's cash position and funding plan, but what matters is setting the rule before you look at the calculation — if you choose the criterion after seeing the numbers, you get the conclusion you want.

Designing a pricing experiment. If you test by splitting the time periods (9,900 won in the first week, 14,900 won in the second week) instead of randomly splitting visitors per price, differences in weekday, ads, and season get mixed into the price effect. Show the prices at random within the same period, and fix them per user so that the same person cannot see a different price by refreshing. Exposing customers who have already paid to a different price loses trust, so pricing experiments usually target new visitors only.

What it looks like in the field

• You set the first price by conversion rate, and the moment costs attach, you find that almost no money is left by each customer.
• The LTV in the investor materials is on a revenue basis and the CAC is a blended CAC, so LTV/CAC looks several times better than it really is.
• You concentrated the budget on the channel with the cheapest CAC, but that channel's customers leave before the payback period.

How to notice when you are wrong

• If the numerator of LTV equals the price, variable costs have been left out. Write down in one line how much smaller the contribution margin is than the price.
• If the CAC table contains channels with 0 ad spend, first distinguish whether it is a blended CAC or a per-channel CAC.
• If the same churn rate was used for every channel's LTV, the differences between channels were erased.
• If the payback period is longer than the average stay of that channel's customers (1/churn), that channel on average loses the customer before getting the CAC back.

What you will do in the next lab

You judge the three price cells by conversion rate, revenue, and contribution margin separately and see how the conclusions split, and calculate the monthly churn rate and LTV from the subscription records. From the ad spend table you work out channel CAC, blended CAC, payback period, and LTV/CAC, and decide price and channel budgets with a rule set in advance. The grader also runs your functions against variant materials.