With 95% certainty, this hidden metric predicts earnings

Iwrite on Medium, and every month it pays money into my bank account. Every month, however, I’m surprised at the amount. I like getting paid, but I wanted to know why.

The answer seems obvious — paying members pay for writers — but I still wanted to prove it from the stats.

So, I collected data from posts that earned money in July, a $270 lifetime value (the actual payout was $200, but that’s just a timing thing). The most vital data was hidden and I had to dig through 496 profile pages to find it, and I did.

These are the results.

Starting Point: Mediums Stats

Medium gives you a pretty nice stats view, but the first thing they show is also the most useless. Views is the standard web metric for advertising — those eyeballs are what you sell — but that’s not how this business model works. I had one post get 5,700 views and earn only $4. Another post only got 300 views but also earned $4.

So, I gathered some data, the views, fans and claps — which I could all get through the normal stats — and the number of ‘member fans’, which I could not. For that I had to manually visit 496 profile pages to see if they said ‘Medium member’. 241 of them were.

Then I ran some basic stats to get a graph for each metric, and a R² to see how much of the variance they explained. A higher percentage means more of the variance is explained by the variable.

R-Squared

Views:        53%
Claps:        76%
Fans:         88%
Member Fans:  95%

As I suspected, the views metric is almost useless. Claps is OK, but also a pain to assemble. The fans metric is much more useful, but still not the best. Engagements by paying members is the best predictor of all.

As you can see, this data fits a linear model very well (R²=95%). The more that paying members engaged with my posts, the more I got paid.

One thing to note, however, is that this data has a significant outlier. One post earned $100. If I exclude that, then the R² is only 83%. Still the strongest effect overall (if I change the other data to match), but not 95%.

In the end, however, I kept the outlier in, because my earnings always have an outlier. There’s almost always one runaway piece each month. In that sense, this is just how the data is, it has a long tail.

The Rule Of Thumb For Earnings

For me this all leads to a rule of thumb — each time a member claps on one of my posts I should earn a dollar. $1.06 to be exact. This number predicts $255 in earnings for my data set, and the real value was $270.

I should note that where I live this amount is half the average household income. Everytime a member claps they’re basically buying me lunch, so I’m thrilled.

What This Means As A Writer

Does this change how I write? Yes. It changes who I write for.

I don’t write for everyone, and I don’t write for the most views. When I sit and write I try to make one thing that will connect to one person. And I’m rewarded for that. I don’t need to attract 1,000 people and hope that five of them are dumb enough to click an ad. I just need to write for one person and hope that they’re smart enough to understand.

I write for the people that value content enough to pay for it, and I try to make content that’s worth their money and time. The Medium model pushes you towards this behavior almost unconsciously, but given that I refresh the stats page 100 times a day I wanted to open a spreadsheet and see for sure.

So now I know, and so do you. If you’re all about the Benjamins, it’s all about the members.

This is my data set. It’s a mess, but have a look if you like. My ‘sample’ was every post that earned money in July. My ‘method’ was to check the stats page for lifetime earnings, check the page for fan counts, and then visit every single profile to see if it said ‘Medium members since’.

📝 Read this story later in Journal.

👩‍💻 Wake up every Sunday morning to the week’s most noteworthy stories in Tech waiting in your inbox. Read the Noteworthy in Tech newsletter.