Example memo · invented company

A decision memo on a price increase

A decision memo answers one question you are about to act on, with the numbers behind the answer, and the analytics director signs it before you act. This example asks whether an invented subscription company should raise its Starter plan from €29 to €31.90 a month. In a three-month test, 203 of 1,200 customers left at the new price against 171 of 1,200 at the old; the 95% range for the extra leaving ends at 5.6 points, below the break-even line of 9.5 points, so the memo recommends the rise. A price test, designed and read out, is a sprint, from €12k, fixed before work starts.

The example · invented company

Decision memo: raise Starter by 10%?

The company is the one in the revenue forecast demo, at its model’s starting point. Its margin, the rise and the test result are invented for this memo, and every figure in it is computed when this page is built.

The question. Raise the Starter plan from €29 to €31.90 a month, 10% more, for the 2,400 customers on it?

The answer. Yes, for existing Starter customers. In a three-month test, 203 of the 1,200 customers at €31.90 left, against 171 of the 1,200 at €29. The 95% range for the extra leaving ends at 5.6 points, below the break-even line of 9.5, so even at that end the rise earns more margin than it loses.

The break-even

Serving a Starter customer costs €5.80 a month, so €29 leaves a margin of 80%, €23.20. The extra €2.90 is all margin: a customer who stays at €31.90 brings €26.10. The rise pays until 11.1% of the paying customers leave because of it, 10% ÷ (80% + 10%): 267 of today’s 2,400.

Over a test, a customer who would have left at €29 anyway pays neither price, so the line applies to the rest. At Starter’s 5.0% a month, 14.3% of a group leaves over three months, and the line is 11.1% of the other 85.7%: 9.5 points of the group.

The test

Starter’s 2,400 customers were split at random: 1,200 moved to €31.90 and 1,200 stayed at €29. New sign-ups kept €29 throughout. The test counted who had left each group after three months. It would have stopped, with the 1,200 returned to €29, had the €31.90 group at any monthly count lost 115 more customers than the €29 group: the line, reached early.

The groups were sized to the line, not to the effect. Showing that the extra leaving sits below the line takes 213 customers a group if the price drives no one away, so 1,200 a group is enough. Had the price driven away half of the line’s share, 4.8 points, 1,200 a group would still have shown it below the line in 8 tests of 10 or more; at three quarters, 7.1 points, it would have taken 4,031. Measuring the effect itself was out of reach: seeing a 2.0-point rise in leaving takes 5,082 customers a group, more than twice Starter’s whole base.

What the test could show with 1,200 customers a group, at 95% confidence and in 8 tests of 10 (invented company)
Question	Customers a group	1,200 enough?
Extra leaving below the line, if the price drives no one away	213	Yes
Extra leaving below the line, if the price drives away 4.8 points	959	Yes
Extra leaving below the line, if the price drives away 7.1 points	4,031	No
A 2.0-point rise in leaving	5,082	No

The result

Leaving after three months, by price (invented company)
Group	Customers	Left	Share left
At €31.90	1,200	203	16.9%
At €29	1,200	171	14.2%
Difference		+32	2.7 points

Extra leaving at €31.90

2.7 pts

16.9% against 14.2% after three months. Invented company.

Upper end of the 95% range

5.6 pts

The range runs from −0.2 to 5.6 points. Invented company.

Break-even line

9.5 pts

11.1% of the 85.8% who stayed at €29. Invented company.

The 95% range runs from −0.2 to 5.6 points. It starts below zero, so the test cannot say whether the new price drove anyone away; it does not have to, because its upper end sits below the line. Applied to all 2,400 Starter customers, three months on, the new price brings €52,043 a month of margin against €47,746 at €29: €4,297 more. At the range’s upper end, the gain is €2,481 a month.

Recommendation

We recommend moving the other 1,029 Starter customers, those still at €29, to €31.90 now. The recommendation rests on three months of leaving. It stops holding if, after those three months, customers at €31.90 keep leaving faster than the 5.0% a month they left at €29. No test at this size can see that: a lasting rise of 1 point in monthly churn would take 8,155 customers a group, more than Starter has. So Starter’s churn is read each month in the monthly revenue report, and a rise there is the point to revisit the price.

Not covered

The price for new sign-ups, which is a separate decision. A test of trial conversion against its own line, 2.8 points below today’s 25.0%, would take 3,815 sign-ups a group: 13 months at 300 a group a month.
Moves up to Growth. At €31.90, the step from Starter to Growth narrows from €70 to €67.10 a month, and the test did not count upgrades.
Whether 10% is the right size of rise, the Growth and Scale prices, packaging, and competitors’ prices.

Method note. Which checks ran on this memo:

Invented inputs: the company, its Starter price, customers and churn are those of the model behind the revenue forecast demo; the 80% margin, the 10% rise, the 2-point change and the test result are made up for this memo. No figure comes from a client.
The €29 group's leavers are its 1,200 customers times the company's three-month leaving at 5.0% a month, rounded to whole customers; the €31.90 group's are 32 more.
Break-even: rise ÷ (margin + rise). Checked on a grid of rises, margins and rates: at the break-even and at the test's line, the two prices earn the same margin.
Group sizes: (1.96 + 0.84)² × (p1(1 − p1) + p2(1 − p2)) ÷ gap², rounded up, at 95% confidence and in 8 tests of 10. Checked against hand-computed cases, this memo's 213 and 5,082 among them, and that the size falls as the gap grows.
The 95% range: the difference ± 1.96 × its standard error. Checked against hand-computed cases, this memo's −0.2 to 5.6 points among them, and that it holds its estimate.
Each direction word in the memo, such as more, below, narrows and enough, is checked against the numbers it describes.
These checks run every time the page is built, and a failure stops the build. Every figure in the memo is computed, none typed in.
Code review: one review agent, working on its own, read this page's code and the module it computes with on 15 October 2026, with the three other pages built on the same company. It found one medium and two minor issues across the four pages; all are fixed, and a second check confirmed each fix.

This memo is an example on an invented company and is never signed, so its stamp stays at “Sign-off pending”.

Sign-off pending · Analytics director

On your data

What you receive for a price test

The same memos, on your own customers instead of an invented company’s.

What it reads. Billing history with each customer’s plan and price, churn history, trial conversion if you sell that way, and the cost of serving a customer, even a rough one. Where the data goes is set out on data handling.

What you receive. A test-design memo before the test, and a read-out memo like this one after it, each of up to two pages, with a method note listing the checks that ran. The price-test calculator shows how large the groups must be on your own rates.

How it is checked. Agents build the analysis and draft the memos. Scripts, not AI, check the numbers and claims, and on what you act on, such as this decision memo, a panel of three reviewer agents and a judge reviews it. Then the analytics director signs it.

Price. A price test, designed and read out, is a sprint, from €12k, fixed in a one-page proposal before work starts. The method behind it is on pricing analysis.

Start here

Planning a price change?

Up to thirty minutes on the change and your data, then a fixed proposal.

Book a free call