A year after Substack, good and bad

I definitely wasn't planning on eating something spoiled and getting food poisoning symptoms right before deadline but here we are, posting thru it 🙃

Time flies, but it's January 3rd of 2024, I had moved off of Substack to the current self-hosted Ghost instance. After almost a year, it seems like a good idea to see how things are going. Hopefully this lets anyone who wants to write have a better sense of what works or not. Writing is scary enough without tech risk laid on top.

Quick recap, why not Substack?

There's already been plenty of discussion of why Substack isn't a good place to be. A short list includes their stance on keeping and even promoting far right wing writers peddling all the hateful nonsense you'd expect. Then they charge a 10% fee which may be fine for small free newsletters but gets expensive as it expands. But finally, they have an extremely aggressive product that tries excessively hard to push user signups and upsells – signs that money spigot isn't flowing as investors want.

So a bunch of people, myself included, has decided it was best to move to something else, especially before they decide to slam the door on their walled hate garden when many alternatives exist now. In my case, I went with self hosting a Ghost instance.

How is self hosting Ghost been after a year?

On the balance, very well. Here's the particular things that are going well:

1. Setup was pretty smooth if you're familiar with the command line. The online instructions are good and for the most part the system runs very well on cheap virtual machines
2. Maintaining the system is also quite good. Updates are easily handled through a built-in script in the provided installer/updater CLI tool
3. The fixed cost of the server is quite cheap, the cost of sending emails is the biggest cost going forward
4. The writing experience works decently well. I can hop between my laptop and my phone, working on the same post, and its autosaving generally avoids edit conflicts unless I abuse it
5. The newsletter does continue to grow without any weird "Substack Organic Growth" stuff. More on that later
6. User engagement still appears to be really strong. Supposedly ~40-50% of every email sent has open tracking data associated, so they're reaching the readers

Here's things that aren't as great

1. Comments and discussion features are pretty sparse and the UX for them is pretty bad
2. Comment MANAGEMENT is garbage. There is no centralized management UI for them in the back end. Luckily the poor UX of comments means there is little to moderate
3. Substack's SEO seems significantly stronger than anything my self hosted domain could muster even after a year of new posts. More on this later too.
4. The handling of unsubscribes and attrition is utilitarian
5. My self promotion skills are very weak, so there's not much net growth in readers

Organic growth w/o Substack does happen, slower

One thing you notice when you have a Substack is that you are constantly getting new people subscribing to your newsletter if you have the slightest bit of content available. More so if another person recommended you. It's probably the biggest reason why some people are afraid to get off the platform, especially if they rely on subscriptions for their sole livelihood.

But in my case, I stopped posting to my old substack on Jan 3rd, 2024. I put a big featured post announcing I was never going to post there again, put links to the new domain prominently on that last post. Then I went back for about a whole year's worth of posts (and the most popular posts in general) and put a header blurb that points people at the new domain in hopes of seeding some site reputation for the new domain.

Despite all the messaging, I STILL got random people occasionally signing up for the mailing list there, because of Substack's hyper-aggressive UI, you'd inevitably get a subscriber who just signs up simply to get past the interstitial placed before they get to even see the existing posts. It'd be a couple of people a week. You'd think that people would read the farewell post and sign up on the new site, but we only saw a fraction.

So, after a year of this nonsense, and getting annoyed at how my old Substack posts continued to have better SEO than my self-hosted domain even after a year, I manually deleted all posts except the last goodbye post. I certainly wasn't going to let someone squat the subdomain either. No trickle of signups is worth the trouble if having old posts on that platform.

But guess what? I still get the occasional email notification that some random person signed up.

In a contextless vacuum, these mechanical subscriptions make Substack seem extremely valuable – hey, look at all these readers "the platform" is bringing! Don't leave! Except, the fact that they keep happening despite my active efforts to discourage them are more indicative that many of these signups aren't particularly engaged readers. They're people just signing up, giving away their email information, so they can see an archive that has nothing.

I don't think the company is doing this right now, but IF one day the VC gravy train collapses for the company and they need to convince more writers to adopt (or remain on) their platform, this avenue would be an ideal way for them to "juice" their stats.

SEO struggles

As I mentioned above, Substack's SEO strength is not to be dismissed. For most of 2024, until I deleted all my posts, if you searched for "Counting Stuff" or my name, Substack was more likely to be the top Google entry than either my self-hosted domain or my personal website that literally uses my name as the domain.

Even if I had identical posts on both sites, Google seemed to consider the old Substack one as canonical and seemed to not even bother indexing some of the new copies. I know there is some way to manipulate canonicity, but it's a pain to do (if it's possible in the platform at all).

Since I honestly don't have any modern SEO knowledge, I'm not exactly sure what is to blame for my generally poor performance, but I assume that it's like 30% my own incompetence, 30% Ghost having 'meh' defaults, and 40% Substack actually hired some people who DO know what they're doing and they built a strong reputation on the search engine.

Deleting all my old posts seems to have helped things a bit, the index seems to be reshuffling now, but I imagine I have a lot more inprovements I could do.

Email bounce attrition and retention

One of the bits of overhead that running your own email newsletter is managing your own email reputation for your domain. If you keep sending email to dead email addresses that bounce emails, or get reported for spam, eventually this gets back and gets you out on a naughty list. Obviously, you do not want this to happen.

Now, Ghost uses Mailgun to manage its email sending feature, and inherits most of its email management features through the Mailgun API. This includes information for when an email has been opened or bounced. If an email bounces a few times, Ghost/Mailgun will automatically disable sending emails to that account in order to protect your email sending reputation. This is generally a good thing, but the management UI makes it hard to surface when this is happening. Early on, I had to check in on the stats every send to make sure things aren't spiking due to misconfiguration or poor subject wording since a new domain sending out thousands of emails gets looked at very suspiciously. Luckily, nowadays, I only have to check things occasionally for anomalies and don't have to think about it much.

Goodbyes are easy

What's not great about Ghost is that there's very little in the way of attempts to convince users to stick around. If they want to unsubscribe, they click a few links, toggle a switch, and they're done. This is also true for any paying subscribers (of which I have precious few), there's no automatic attempts at some customer retention, discounts, etc., there's not even a short feedback collection box. I know some people feel such efforts are essentially "dark patterns" but as a long-time SaaS analyst, I know that there's room to at least make an attempt at a respectful retention prompt without being evil or deceptive. The lack of anything is a bit of a gripe when for a small newsletter like mine, every subscriber counts towards paying the bills (there's only about ~30ish people w/ an active paid subscription so far).

All told, since I'm not very aggressive about promoting signups, the newsletter has essentially held ground this year. We've gained a bunch of readers, but at the same time attrition also got rid of a similar number. Maybe I just need to be much better about self-promotion too.

Let's hope for a less eventful 2025

I've got a bunch of little ideas in my head for next year, but the hope is that im going to finish moving into my house and finally be done with stressful renovations. So I'm hoping for a less exciting 2025.

At least with respect to my writing. The political climate is very likely to be very... not.

Standing offer: If you created something and would like me to review or share it w/ the data community — just email me by replying to the newsletter emails.

Guest posts: If you’re interested in writing something, a data-related post to either show off work, share an experience, or want help coming up with a topic, please contact me. You don’t need any special credentials or credibility to do so.

"Data People Writing Stuff" webring: Welcomes anyone with a personal site/blog/newsletter/book/etc that is relevant to the data community.

About this newsletter

I’m Randy Au, Quantitative UX researcher, former data analyst, and general-purpose data and tech nerd. Counting Stuff is a weekly newsletter about the less-than-sexy aspects of data science, UX research and tech. With some excursions into other fun topics.

All photos/drawings used are taken/created by Randy unless otherwise credited.

• randyau.com — Curated archive of evergreen posts. Under re-construction thanks to *waves at everything

Supporting the newsletter

All Tuesday posts to Counting Stuff are always free. The newsletter is self hosted. Support from subscribers is what makes everything possible. If you love the content, consider doing any of the following ways to support the newsletter:

• Consider a paid subscription – the self-hosted server/email infra is 100% funded via subscriptions
• Send a one time tip (feel free to change the amount)
• Share posts you like with other people!
• Join the Approaching Significance Discord — where data folk hang out and can talk a bit about data, and a bit about everything else. Randy moderates the discord. We keep a chill vibe.
• Get merch! If shirts and stickers are more your style — There’s a survivorship bias shirt!