With your data

Should we raise the price for new subscribers?

A founder is about to move new subscribers from 9 to 12 dollars. She tried 11 once and sign-ups dropped, but she never counted. Her own subscription export settles what that test really showed.

Everything in the grey boxes is what Learning Nodes returned in real runs, selected and unedited. The app and its subscription file are made up for the example.

INTERACTIVE VERSION · TWO MINUTES

Click through a founder's session in the Workstation, one step at a time. A short note explains each screen. No sign-up, and nothing is saved.

The situation, as typed

I sell a meal planning app at 9 dollars a month and I am thinking of moving new subscribers to 12 dollars a month from November. Existing subscribers would stay on 9 for a year. About 1,400 people pay today, around 180 new ones join each month, and roughly 60 cancel. I have an export of every subscription since launch with the start date, the end date if they cancelled, the plan price they joined on, the country, and how many days they stayed. Last year I tried 11 dollars for six weeks and sign-ups dropped, though I never counted by how much. I want to know how many of the people who joined in the 11-dollar weeks stayed compared with the 9-dollar months, and whether cancellations bunch up in the first month.

The decision Learning Nodes proposed

Decide whether to move new subscribers to 12 dollars a month from November

Move new subscribers to 12 dollars a month from November
Keep new subscribers at 9 dollars a month

You are deciding whether to move new subscribers to 12 dollars a month from November or keep them at 9 dollars a month. It turns on how much sign-ups dropped in the past 11-dollar test and whether people who joined at the higher price stayed as long as those at 9 dollars.

What would make each option right

Move new subscribers to 12 dollars a month from November

The drop in sign-ups at 11 dollars was smaller than the extra revenue from new subscribers at 12 dollars, and people who joined at the higher price stayed at least as long as those at 9 dollars.

Keep new subscribers at 9 dollars a month

Sign-ups dropped by more than the extra revenue from new subscribers at 12 dollars, or people who joined at the higher price cancelled sooner than those at 9 dollars.

The questions in the way

How many of the people who joined in the 11-dollar weeks stayed compared with the 9-dollar months
Whether cancellations bunch up in the first month
How many sign-ups per week occurred in the six weeks at 11 dollars compared with the surrounding weeks at 9 dollars?

Claims nobody had checked

Last year I tried 11 dollars for six weeks and sign-ups dropped, though I never counted by how much

The case for keeping the price at 9 dollars rests on this unmeasured drop.

Cheapest test: Count sign-ups per week in your subscriptions_since_launch file for the six weeks at 11 dollars and the weeks at 9 dollars around them.

Existing subscribers would stay on 9 for a year

The revenue from the price increase depends on this one-year hold for existing subscribers.

Cheapest test: Check your own records for how long you have kept existing subscribers at the old price in past changes.

Answers from the data

She attached the subscription export and pressed Answer from data. Learning Nodes wrote the queries from the column names, ran them in her browser, and answered with the numbers next to the queries that produced them. It also said plainly what the file could not tell her.

How many of the people who joined in the 11-dollar weeks stayed compared with the 9-dollar months, and whether cancellations bunch up in the first month

Of 1000 total subscribers 860 joined the 9-dollar plan and 140 joined the 11-dollar plan. You see 684 of the 860 nine-dollar subscribers remain active. You see 86 of the 140 eleven-dollar subscribers remain active. Days active values are missing in 77 percent of rows so first-month cancellation shares rest on very few records. The file supplies no information on exact join dates or reasons for cancellation.

SELECT plan_price, COUNT(*) AS subscriber_count, SUM(CASE WHEN status = 'active' THEN 1 ELSE 0 END) AS active_count FROM data GROUP BY plan_price
SELECT plan_price, CASE WHEN days_active < 30 THEN 'first_month' ELSE 'later' END AS cancel_timing, COUNT(*) AS cancelled_count FROM data WHERE status = 'cancelled' GROUP BY plan_price, CASE WHEN days_active < 30 THEN 'first_month' ELSE 'later' END ORDER BY plan_price, cancelled_count DESC LIMIT 10
SELECT plan_price, COUNT(*) AS total_cancelled FROM data WHERE status = 'cancelled' GROUP BY plan_price

subscriptions_since_launch.csv, 1,000 rows

Bring your own file: attach a CSV to any situation and press Answer from data.

What the founder committed (illustrative)

Keep new subscribers at 9 dollars through November and test 12 dollars on one country first

Founder

684 of 860 subscribers who joined at 9 dollars are still active
86 of 140 subscribers who joined at 11 dollars are still active
The sign-up drop during the 11-dollar weeks is still uncounted, so it becomes the next question
First-month cancellation figures rest on too few records to lean on

Frozen when committed. The one-country test is a new situation that starts from this record.

See the same case in the product, one step at a time.

She came in ready to raise the price. Her own file showed that people who joined at the higher price were leaving faster, and it showed which number she still had to count before deciding for good.

Book a demo