History of This Newsletter In One Chart
With a system design lesson at the end
The recent price increase announcement, the flurry of signups that resulted, & now the aftermath has got me thinking about this newsletter as a business. Yes, I know this isn’t software design, but we’ve established in book 1 that economics profoundly affects software design, so I don’t mind thinking about it for a moment. I’ll get back to coupling & cohesion & power laws in a minute.
In reviewing the revenue per day for this newsletter, I realized that I can tell the story from a single graph. Here it is:
This is the revenue per day for every day of the 2 years I’ve offered paying subscriptions. I’ll annotate it with significant events:
Initial signups. The day I announced paying subscriptions the quantity of signups weas significant, even on the scale of later signups. I interpret this as me having enough of a reputation that people were willing to take a chance on paying and also that I had been publishing regularly before that. To put the numbers in context, 1.8% of my total subscribers converted in one day. (Most newsletters have between 1-5% paying subscribers.)
The dark blue spike below that initial day is the second payment of the $70/year fee for all those zero day customers. Approximately 90% of all paying subscribers pay annually, so revenue is never going to be smooth month-to-month.
The “productivity” spike is around the publication of the 2 “No, you can’t actually measure programmer productivity” posts. The 2 were published a week apart, but the first one resulted in by far the most sign-ups.
Then we come to the January 3 announcement of the impending price increase. That was 3X any single day I’d done previously. My total subscriber base had grown enormously, so that represented 0.3% of total subscribers beginning to pay.
Finally, the day of the price increase outpaced even the announcement day. I can tell you that was an exciting day, watching the subscriptions roll in minute by minute.
Annual renewal deserves a mention. When payment is annual, I see significant churn on the renewal day.
Generally my churn is around 5% per month, which is normal for newsletters. Around the first anniversary, though, churn bounced up to double that. I wasn’t doing anything different, it was just that a different segment of readers signed up in that initial cohort.
Deming teaches treating special case variation differently from common case variation. If you react systemically to every random variation, you’ll thrash. If you treat every systemic problem as a one-off, you’ll never address your real bottlenecks. That first year churn is special case variation. There’s nothing I can or should do in response. If overall churn starts climbing, though, then I need to react.
See, I slid a system design lesson in regardless of topic.
In the publishing industry, which you're certainly a part of with this substack, the rule for measuring gross *net* income (rather than gross income, which is what you're counting) is to distribute the yearly subscription dollars over the duration of the subscription. To wit, if you promise 52 weeks of content for an annual subscription, after week 1 only 2% of the dollars you received from that subscription are actually yours. The rest represent a liability you have to that customer, which is reduced with each week.
It's a helpful distinction to bear in mind, so that your budgeting correctly reflects what dollars are truly yours--a number that will change every month, to the extent that you have annual subscriptions.
Your post underlines the importance of having a paid launch strategy that is well thought out. Even Lenny Rachitsky wrote about this and how he saw a significant rate of conversion on paid launch day. The other takeaway is creating significant FOMO can increase sign ups too! Thanks for sharing.