Substack Notes Analytics: What You Can See and What You Can't

Substack does give you analytics for your notes. You can open a note, find its stats, and see how it did. That part works.

The problem is the next question. You want to know whether image notes beat text notes for you. Whether Wednesday is hurting you. Which topic earns comments rather than just likes.

Those questions need you to look across many notes at once, and that is exactly what you cannot do.

What Substack actually shows you

The per-note stats are real and useful. You reach them from your profile's activity area, and the mobile app tends to show the clearest numbers, so if you have only ever used a browser, try the app.

You get reactions, restacks and comment counts for a single note. The reporting window covers roughly the last three months rather than your full history, which matters if you post often.

One caution: do not trust any screenshot in any article, including this one. Substack ships changes regularly. Check your own account for what is actually there.

You can see one note, not the pattern

Every useful question about your notes requires grouping, and grouping is the part that is missing.

You can open a note and see it got two reactions. You cannot ask which format averages best. You cannot filter by weekday. You cannot compare topics. Each of those needs a table, not a single row.

You can see one note. You cannot see what your notes have in common.

That is not a small gap. It is the difference between knowing a result and knowing what caused it.

Wednesday was my worst day, and also my busiest

Here is what I found pulling my own 933 notes into a spreadsheet. The baseline average is 0.9 reactions per note. Now look at what moves it.

By format:

• Image notes: 1.4 average, across 29 notes
• Link notes: 1.0, across 24 notes
• Text only: 0.8, across 880 notes

By hour posted, 21:00, 20:00 and 12:00 all reach 1.1. Those are my three best windows.

By weekday, Wednesday sits at 0.7, my worst. It is also the day I posted 156 notes, more than any other. I had been doing the most work on my worst day without knowing it.

None of that is visible one note at a time.

The category number that looks best and means least

By topic, Culture scores 3.1 average reactions. Technology scores 1.3 across 173 notes.

Culture looks like my best category by a wide margin. It is also my least reliable number. That 3.1 comes from only 12 notes, and one or two that happened to land well can drag an average that small a long way.

Technology at 1.3 is built on 173 notes. That one I can trust.

A high average from 12 notes is not a finding. It is a guess with good lighting.

Always check how many notes sit behind an average before you act on it.

Reactions are the weakest of the three signals

I used reactions in those averages because they are the most common data point. They are also the cheapest to give. Clicking a heart costs nothing.

A restack costs the other writer something: they are putting your note in front of their own audience under their name. I have 83 restacks across 933 notes.

A comment costs the most. Someone stopped scrolling and typed.

Keep all three in separate columns and never add them into one score. A note with no reactions and two comments is doing better than one with five reactions and silence.

How to do this with no tool at all

Open a spreadsheet, one row per note, with these columns: date, hour, weekday, format, topic, then reactions, restacks and comments kept separate.

Fill it in as you go rather than retroactively. Going back through hundreds of notes by hand is not a good afternoon.

Once you have fifty or so rows you can start filtering. Sort by format. Group by weekday. Average by topic. The patterns show up faster than you would expect.

The only real cost is discipline. You have to log each note when you post it, and if you skip a fortnight the gaps make your averages unreliable.

Or have it kept for you

That discipline is the part StackBuddy removes. It keeps the history automatically as each note goes out, and prints the note count beside every average, so a number built on twelve notes cannot quietly pass itself off as a finding.

It also connects to AI tools over MCP, so you can ask across your full history in a conversation rather than building filters: which format should I use this week, what time works best for text notes, what earns comments rather than likes.

Every account starts with 14 days of Pro and no credit card, and analytics is part of Pro, so you can look at your own history before deciding whether it is worth paying for.

You still need to understand what the numbers mean before handing them to any tool. That is what this article is for. But once you do, having the history kept and queryable saves the genuinely tedious part.