The Substack Notes Algorithm, Explained

Published: September 2026

Most "algorithm explained" posts are guesswork extrapolated from a handful of viral Notes. This one has an actual advantage: Substack's own team has directly explained what the algorithm optimizes for, on the record, at a real event — which means the honest answer here doesn't have to be a theory.

What Substack itself says it optimizes for

At Substack's NYC Notes event in November 2025, Mike Cohen (Substack's Head of Machine Learning) put it plainly: "The goal is to get people to discover, subscribe, and ideally pay" (reported by Wes Pearce, via Escape the Cubicle). Co-founder Hamish McKenzie framed the business logic behind it directly: "We want you to grow... make as much money as possible. We only make money when you make money."

That single sentence explains a lot of behavior that otherwise looks inconsistent. The algorithm isn't optimizing for time-in-app or ad impressions the way most social feeds do — it's optimizing for you converting a reader into a subscriber, because Substack's own revenue depends on that happening.

How it actually decides who sees your Notes

Cohen described the mechanism directly: "We take a look at who you are... what things you're subscribed to, who you follow, and what interests you've specified... We try to turn that into a numerical representation." In practice, that means the algorithm builds a profile from your subscriptions, your follows, and your declared interests, then matches that profile against writers with overlapping signals — along with more basic context like where you are and what language you use.

The second piece is audience overlap: "If we see overlaps between audiences of different publications, that becomes a virtuous cycle that feeds back." When your Notes get engagement (especially restacks) from accounts whose audience overlaps with yours, the algorithm treats that as a signal that your Notes are worth showing to more of that same audience — which is the actual mechanism behind why engaging with other creators drives discovery, not just posting more yourself.

The trust mechanism, and why it resets

Beyond matching interests, the algorithm appears to build an ongoing trust profile per account, based on consistency, engagement rate, restack behavior, and how an audience interacts with an account's Notes over time. The specific detail worth knowing: posting daily for a week and then going quiet for ten days reportedly resets whatever trust the algorithm had started building — meaning the benefit of a consistent posting streak isn't just visible momentum, it's also not accumulating a kind of invisible credit the algorithm is tracking underneath. This is a real, mechanical reason posting consistency matters more than any single viral Note — sporadic bursts don't just look less professional, they appear to genuinely reset a signal the algorithm is using.

Engagement velocity in the first 30 minutes

A specific, measurable pattern: the algorithm reads engagement velocity in the first 30 minutes after a Note goes out, and uses that early signal to decide how far to push it. This is also why promotional Notes reportedly underperform conversational ones by 3–5x — they tend to generate less immediate reaction in that critical early window, which caps their distribution before they've had a chance to build momentum any other way.

The links nuance: not all links are treated the same

Earlier research on this blog noted that linking to your own Substack posts doesn't suppress reach the way it would on platforms designed to keep you scrolling in-app. That holds specifically for links that keep a reader inside Substack's own ecosystem. Links out to external platforms are a different story — Notes containing them have been reported to see sharply reduced views (in some comparisons, around 20 views versus 1,000+ for a comparable Note without an external link). The likely explanation lines up with everything else here: Substack's algorithm is optimized to convert readers into subscribers on Substack, and a link sending someone off-platform works against that specific goal in a way that a link to your own Substack post doesn't.

What this explains about format

Because the goal is genuinely to move readers toward subscribing (not to keep them scrolling inside Notes indefinitely), the algorithm's incentive structure is the opposite of most social feeds in one specific way — but it's still actively working against anything that takes a reader off Substack entirely, which is worth remembering before dropping an external link into a Note for reach's sake.

What it doesn't reward

Notably absent from Substack's own explanation: any special reward for a specific format, posting time, or Note length. The mechanism is about audience-overlap, consistency, and subscription intent, not a magic formula — which lines up with why format and timing research keeps finding audience-specific answers rather than one universal trick that works for everyone.

The practical takeaway

Given the algorithm is explicitly rewarding audience overlap, consistency, and subscription intent, the highest-leverage actions are the ones that build real overlap and keep the trust signal intact: restacking and replying to writers whose readers would genuinely like your Notes, posting consistently enough not to reset whatever trust has built up, and keeping links pointed at your own Substack rather than off-platform. Posting volume and format still matter for getting seen at all — but per Substack's own team, the actual ranking signal is about connection and consistency, not raw output.

See exactly which of your Notes are building that audience overlap, and track your own posting consistency over time. NotesIQ tracks engagement and growth from your own real data, not a guess about the algorithm.