The Newsletter Metrics That Actually Predict Revenue
Most creators chase likes, opens, and subscriber counts. The data shows clicks, replies, repeat buyers, and referrals are what really compound.
You’ve been tracking the wrong numbers.
Not because you’re not paying attention. Because the platform makes it easy to see the metrics that feel good and hard to find the ones that actually predict growth.
Numbers like opens, likes, subscribers, tell you what happened. They don’t tell you what’s compounding.
There’s a different set of metrics underneath those. The ones that predict whether a reader becomes a buyer, whether a buyer comes back, and whether your audience is sending you new customers while you sleep.
Finn Tropy has spent years building analytics tools for Substack creators and analyzing data across thousands of newsletters. I’ve spent 15 years tracking the revenue metrics that tell you whether a sales system is working.
We looked at both sets of data and found the same thing: the metrics most creators ignore are the ones that actually compound.
Today we are going to walk through the metrics that you should be tracking and show the data that supports them.
I’ll give you the metrics that I track and Finn uses the Data to hammer home the point. Buckle up because this article is an engineer’s delight. We have all the numbers. 😁
Let’s drive in!
The Vanity Metric Trap
These are the metrics that are noise in your growth system. That’s not an opinion. It’s what the data shows and it’s what I see in every sales pipeline I’ve ever audited.
A like costs nothing. It takes half a second. It generates a small dopamine hit for the creator and zero commitment from the reader.
A restack with commentary costs something. It required the reader to form an opinion, write a sentence, and put their name on an endorsement. That’s a signal that matters.
A DM costs even more. Someone had to stop scrolling, open a conversation, and type a message to a stranger. That’s a desire to start a relationship.
In sales, we have the same distinction. An email open is a like. A reply is a restack. A booked call is a DM. The first two feel like progress, but only the third one is actually progress.
The creators who compound aren’t optimizing for the metrics that are easy to see. They’re paying attention to the ones that require effort from their reader.
Effort is the signal. Passive engagement is just noise.
Customer Engagement Metrics: What Actually Signals a Real Relationship
The Customer Engagement engine of mine Revenue Flywheel System runs on one discipline: knowing who is actually engaged versus who is just present.
In a sales context, an engaged customer is one who:
Replies to your communications with something specific
Asks questions that reveal they’ve been thinking about your work
Refers someone without being asked
In a newsletter context, the parallel signals are:
Comments on your article — not just opens it
Restacks with commentary — not just passive restacks
DMs that reference something specific you wrote
The metric to track isn’t “how many people saw this.”
It’s “how many people did something because of this.”
That distinction is where the flywheel either starts spinning or stays still.
📊 FINN’S DATA SECTION — Customer Engagement
I ran this on 2,204 mature Notes (7+ days old) joined to Substack Notes metrics, plus 1,502 free subscribers and 776 Gumroad buyers on finntropy.substack.com.
The surprise: likes barely predict anything.
On my Notes, Pearson correlation with free-subscriber conversion is:
David’s instinct is right. Likes are noise. In my dataset, Notes that actually converted had lower average reaction counts (14 likes) than Notes that didn’t convert (47 likes).
Virality and conversion are not the same game.
What does correlate with real relationships?
Effort signals, but you have to read them as rates, not raw counts:
Notes with at least one reply: 9.2% converted to at least one free sub vs 6.3% baseline across all Notes.
Notes with at least one restack: 9.2% converted.
Notes in the top reply decile: averaged 0.45 free subs per Note vs 0.03 in the bottom decile. That’s a 14× gap.
Notes in the top click decile: averaged 0.47 free subs per Note vs 0.04 in the bottom decile. That’s a 11.8× gap
Subscriber-level engagement (Gumroad buyers vs everyone else):
Among subscribers I could match to Gumroad purchases (185 buyers vs 1,317 non-buyers):
Buyers comment 2.7× more and click 4.3× more than subscribers who never purchased. Opens move too, but the gap is smaller. Opens feel like progress.
Clicks and comments look like intent.
A “healthy” engagement ratio to watch: on Notes, I track clicks per 100 reactions. My top-converting Notes (5+ free subs) combine high replies + high clicks, not high likes alone.
Two live examples:
18 free subs · 212 clicks · 29 replies — “I analyzed 1.78 million Notes…” (131 likes, 27 restacks)
12 free subs · 157 clicks · 29 replies — “I scraped 3.8 million Substack posts…” (130 likes, 9 restacks)
The pattern is effort + click-through, not applause.
The metric that surprised me most: total clicks on Notes. I expected engagement score (likes + restacks + replies) to predict subscriber growth. It doesn’t not in the aggregate. Clicks do.
David’s “DM = booked call” analogy holds: the action that costs the reader something meaningful (clicking through, replying, DMing) is the signal.
Likes are the email open of Substack.
Improvement Metrics: Which Content Signals Predict What to Build Next
The Improvement engine is where most creators leave the most value on the table.
They publish. They check the dashboard. They see opens and likes. They start writing the next piece without ever asking: what did this post tell me about what my audience actually needs?
The improvement metric is a signal number.
The signals that I have seen predict what to build next:
Reply rate on specific topics — not your overall reply rate. Which posts generated replies where readers described a specific problem? That topic cluster is your next product or service. This is a great insight to use with our Notes. If you have a Note that drives a lot of conversation look to expand that into an article.
DMs triggered by a single piece — if one post generates five DMs in a week, something in that post named a problem specifically enough that five people felt compelled to reach out. That’s a market signal you want to lean into.
Restacks with commentary — when someone restacks your post and adds their own perspective, they’re telling you the post gave them something worth building on. What did those posts have in common? That’s your next 90 days of content.
The improvement driver turns your content calendar into a product roadmap. But only if you’re reading the right signals.
📊 FINN’S DATA SECTION — Improvement Metrics
Which content signals most reliably predict subscriber growth?
Across 2,204 mature Notes, only 6.3% (139 Notes) drove at least one free subscriber. Total free subs attributed to Notes: 230.
Growth is concentrated. Most Notes do nothing; a few do a lot.
The strongest leading indicator in my data is click decile, not like decile:
Top-decile click Notes average 13× more free subs than bottom-decile. When I’m deciding what to expand into a long-form post, I look at Notes that generated replies + clicks in the same week — not Notes that only collected likes.
Reply rate and conversion, is there a threshold?
Notes with zero replies average 0.05 free subs. Notes with one or more replies average 0.19 — a 3.6× lift.
There isn’t a magic number like “5 replies = product-market fit,” but once a Note crosses into the top reply decile, conversion rate jumps to 15% of Notes in that bucket (33 of 220 Notes).
Topic clusters from my top converters (Notes with 5+ free subs):
Data/analysis posts — e.g. 1.78 million Notes (18 subs, 212 clicks) and 3.8 million posts / 1,600-word sweet spot (12 subs, 157 clicks)
Contrarian positioning (”Everyone says validate the market first. I didn’t.”) — fewer likes, strong reply threads
The improvement signal isn’t “this got popular.” It’s “this made people talk back or click through.” Those are your product and post candidates.
A quick note from David here: This is why you focus on the one problem you can solve. Niching down to that one problem helps you find the right audience, that will drive engagement. Now back to Finn.
Pro Studio surprise: the metric that most changed my revenue wasn’t subscriber count — it was repeat purchase rate on Gumroad. I built Pro Studio to solve my own scheduling problem. The data showed that buyers who engaged before purchasing (comments, clicks, DMs) came back. Improvement isn’t just “make better content.” It’s “make the customer better every week” — and measure whether they return.
We all want to be better sales people… the good news is we have the tools we need to win all the deal. Thanks for joining the discussion today. Don’t forget to subscribe and get your free Revenue Flywheel Assessment tool
Re-Engagement and Referral Metrics: The Compounding Case
This is the section most creators skip entirely and it’s where the real revenue multiplier lives.
Two facts that should change how you think about your existing audience:
Fact 1: Maintaining a customer is five times less expensive than acquiring a new one. In sales, this is foundational. In creator businesses, almost nobody tracks it.
Fact 2: Your most engaged existing customers are your best acquisition channel. The referral that comes from a satisfied reader converts at a higher rate than any cold outreach you’ll ever do because the trust is pre-built.
The re-engagement metrics that matter:
Repeat purchase rate — what percentage of your revenue comes from people who have bought from you more than once? If you don’t know this number, you don’t know whether your business is compounding or just acquiring.
Referral source tracking — how many of your new subscribers or customers came from an existing reader’s recommendation? This is the flywheel’s highest-leverage metric and the one almost nobody tracks.
Re-engagement response rate — when you reach back out to a past buyer or inactive subscriber with something useful (not a pitch), what percentage respond? That number tells you whether your existing relationships have value you haven’t activated yet.
The re-engagement metric isn’t about getting more from your audience. It’s about recognizing the value that’s already there and showing up consistently enough to compound it.
📊 FINN’S DATA SECTION — Re-Engagement and Referral
Repeat purchase rate (Gumroad, all-time, paid non-refunded):
More than one in five buyers come back. And they generate 41 cents of every revenue dollar. That’s the compounding case David describes — not hypothetical, just rarely tracked.
Engagement depth → repeat purchase:
Repeat buyers aren’t passive readers. They comment more, click more, and show up more often. Re-engagement isn’t a email tactic — it’s a relationship depth metric.
Upsell flows (re-engagement in practice):
These are automated re-engagement paths — past buyers seeing the next thing they need. The Notes Scheduler → Pro Studio path alone generated $924. That’s the flywheel’s third driver running without manual outreach.
Referral / acquisition source data:
Free subscriber attribution (1,501 free subs with attribution data):
Notes are my #1 free-sub acquisition channel — not email forwards, not SEO alone. That’s reader-driven distribution compounding in public.
Gumroad purchase referrers (top channels by revenue):
Roughly 36% of Gumroad revenue traces back to my own Substack + Gumroad properties. Substack isn’t just audience — it’s a sales channel. Creators who treat it that way track referrer data. Most don’t.
Bridge metric (engagement → purchase): among subscribers matched to Gumroad, 58 of 185 buyers (31%) had left at least one comment before purchasing. DMs are harder to quantify at scale, but my highest-LTV relationships (repeat buyers, collaborators, affiliates) almost all started with a specific reply or DM — not a like.
The Revenue Flywheel Metric Stack
Put the three sets of metrics together and you have a system.
Customer Engagement metrics tell you who is actually in a relationship with you, not just watching. Track reply rate, DMs per post, and restacks with commentary.
Improvement metrics tell you what your audience is trying to tell you they need. Track reply topics, DM triggers, and restack commentary themes. These are your next product features and your next 90 days of content.
Upsell metrics tell you whether your existing relationships are compounding. Track repeat purchase rate, referral source, and re-engagement response rate. These numbers tell you whether you’re building a business or just an audience.
Most creators track the first column on their dashboard. The ones who compound track the signals underneath it.
The flywheel is about seeing more clearly what’s already working and doing more of that.
Action Step
This week, pull your last 10 articles.
For each one, write down:
Did this generate a reply, a DM, or a restack with commentary?
Circle the three that generated the most of those.
That’s your signal cluster and your next 90 days of content direction.
Finn’s move:
If you’re on Substack and can only turn on one metric this week: track Note clicks, not likes.
In Substack’s Notes analytics, clicks show intent — someone left the feed and went somewhere (your profile, a link, a subscribe flow). In my data, clicks correlate with free-subscriber conversion at r = 0.77. Likes correlate at r = 0.003.
Practical first step:
Open your last 20 Notes in Substack analytics (or export via StackContacts / Substack Pro Studio if you have them).
Sort by clicks, not reactions.
For the top 5, write one sentence: What problem did this name specifically enough that someone clicked?
That’s your next week’s content cluster.
Second metric to add once clicks are habit: reply count per Note. When replies and clicks show up together, you’re looking at Improvement signals — not vanity.
Closing
David Roy writes Eng Sales — a weekly newsletter for technical founders and small B2B team leaders who’ve defaulted into the head of sales role. The Revenue Flywheel System is the three-driver framework he uses to turn early customer relationships into compounding revenue. Find him at engsales.substack.com.
Finn Tropy writes Finn’sights — I study creator growth the way engineers study systems—with experiments, dashboards, and curiosity. I build tools, test ideas on myself, and share what truly moves the needle for busy Substack creators.
He also has a great Gumroad store!