Market, price and trust · Pricing

Per seat or per use

Cost a skewed usage distribution band by band, choose between a flat seat price, per-use pricing and an allowance, and show the margin each gives.

casemediumExercise · 3 min · 50 of 59

On this page

The drafting add-on loses money on its heaviest seats and overcharges its lightest, and two teams have two fixes. The pricing lesson pulled the cost levers; this is the charging model.

The situation

The drafting add-on sells at $5 a seat a month to 12,000 seats, $60,000 of revenue. After the output cap and the router from module 4, the token bill is about $27,500 a month on 200,000 drafts a day: assume $0.0046 a draft and a 30-day month. The cached-prefix saving is not counted, because nobody has read the vendor's rate. Gross margin is 54%, and finance holds add-ons to 70%.

Usage is skewed: the median seat drafts 10 a day, the p95 seat 60, the mean about 17. In four bands, 6,000 seats average 6 drafts a day, 3,600 average 16, 1,800 average 35 and 600 average 72. The two heavy bands are the best-fit customers, teams that carry tickets overnight. Assume a draft saves an agent about a minute.

Sales proposes $7.65 a seat, flat. Finance proposes 1.5 cents a draft and no seat fee. Renewals open next month.

The question

Pick the charging model for next year's price list — a flat seat price, per use, or something between — and show the gross margin it gives and whose bill changes.

A strong answer · Rubric · Where answers go wrong

Write your answer first, then open

A strong answer

Per band at $0.0046 × 30 a month: light seats cost $0.83, mid $2.21, heavy $4.83, top $9.94. At $5 the margins are 83%, 56%, 3% and −99%. The light half pays for the tail, which is why the mean sits so far above the median.

Flat $7.65 reaches 70% on paper, $91,800 against $27,500, by charging the light half 53% more for nothing new. They are the seats most likely to drop an add-on they barely use: if 3,000 leave, revenue falls $22,950 and cost only $2,480, and the margin is back to 64%.

Per use at 1.5 cents: about 6 million drafts a month is $89,900, a 69% margin. But the heavy seat's bill goes from $5 to $15.75 and the top seat's to $32.40, different every month, and every draft now has a price, so admins ration the short tickets the router made cheap. It taxes the best-fit customers hardest.

Decision: keep $5 with 500 drafts a seat a month included, and sell blocks of 10,000 drafts at $150, the same 1.5 cents. Light and mid seats see no change. The heavy band adds 550 × $0.015 = $8.25 a seat and the top band $24.90: $14,850 + $14,940 = $29,790 on top of $60,000, a 69% margin. A heavy seat pays $13.25 a month for about 35 agent-minutes a day.

What would reverse it: drafts per heavy seat falling after the first block invoice, which would mean the price is rationing the feature where it fits best; then raise the allowance, not the price. The cached-prefix rate, once read, adds about 5 points to any of the three.

Rubric

Costs each band at $0.0046 × 30 and shows the margin per band at $5, including the negative one.
Reads the mean of 17 against the median of 10 as the tail setting the price.
Tests the flat raise against the light band leaving, with the revenue and the cost each lost seat takes with it.
Prices per use and says what it does to the heavy seats' bills and to how the short tickets get used.
Picks a model, shows its margin, and names which bands see their bill change.
Names the evidence that reverses the choice, and keeps the unread cache rate out of the sum.

Where answers go wrong

"Flat $7.65: 70% is 70%." It sets the price on the mean and charges the 6,000 lightest seats for the 600 heaviest. It never asks what the light half does at renewal, and losing half of them puts the margin back at 64% with fewer customers.
"Per use, so revenue tracks cost." The margin is right and the incentive is wrong: every draft gets a price, admins ration the cheapest drafts first, and the best-fit customers' bills triple. A reviewer reads it as pricing for finance against the product.
"Keep $5 and let caching fix the margin." It counts a rate nobody has read, and the 600 top seats lose about $5 a month each today and still about $3 if the cache takes 18% off the bill.