Your best-performing post probably made you nothing.

That is the uncomfortable finding inside a new data study from Orel Zilberman's The Substack Data Analysis, which pulled every original Note posted between January 1 and September 18, 2026 by 1,569 writers who connected their Substack stats — 396,143 Notes in all, each one carrying impressions, clicks, free subscribers, paid subscribers and the annual revenue those paid subscribers are worth.

The headline split: 0.70% of Notes sold one or more paid subscriptions, against 13.1% that brought in one or more free subscribers. The 2,781 that did sell moved $1,027,701 in paid subscriptions.

Everything else — the likes, the restacks, the screenshot-worthy ones — produced audience and nothing more.

The overlap test is the part that stings

The obvious assumption is that paying readers are just free readers who went further. So the study tested it directly: it ranked every Note by free subscribers gained, took the top 2,781 (the same number as the paid sellers, which works out to Notes with six or more free subscribers), and checked how many sold a paid subscription.

16.5%. The other 83.5% of the year's best free-subscriber Notes sold nothing.

Read that as a coach and it stops being a Substack story. The Reel that added 400 followers and the caption that booked a discovery call are two different products. Most people only ever make one of them, then wonder why the audience number keeps climbing and the calendar doesn't.

Reach is not the variable

The scale numbers are brutal in a useful way: across all Notes, 33 paid subscribers per million impressions, or about 2.6 per 10,000 clicks, with each paid subscriber on a typical selling Note worth $99 a year.

And going bigger doesn't fix it. A companion study from the same dataset found that paid conversion peaks at a few hundred clicks and collapses on viral Notes — 2.89 per 10,000 clicks in the 200-to-999-click band, falling to 0.45 above 5,000 clicks.

The wide net catches browsers. The small one catches buyers.

What the selling posts actually contained

Same dataset, earlier cut, comparing Notes that converted a paying reader against Notes that only converted free ones:

• They linked to a specific named post nearly twice as often: 18.7% versus 9.9%.
• They named a price or a discount at 10.7% to 12.6%, against 6.4% for free-only Notes, and contained a number more often (44% to 48% versus 38%).

Named thing. Real number. Stated price. Direct link. It is almost rude how unglamorous that list is.

The concentration problem

One more figure worth sitting with. Only 290 of the 1,569 writers sold a single paid subscription from a Note in that window, 81.6% of paid-selling Notes sold exactly one, and the top tenth of those 290 writers account for 71.5% of all paid subscribers.

So this isn't a platform where selling happens quietly in the background for everyone. It's a platform where a small group deliberately writes the second product, and almost nobody else does.

The fix costs nothing. Open your own analytics, list the posts that produced actual revenue this year, and put the like counts next to them. The gap is the whole lesson.