5 Years of Newslettering

January 19th, 2021, will make 5 years of sending my NYC Data Jobs Newsletter.

Sending out a newsletter has been one of the most rewarding things I’ve done over the last few years. In the hopes that talking about my experience would be helpful to people who are either just starting out, or are thinking about it (and, selfishly, encourage people to write more good newsletters that I can read), I’ve written down a few observations and experiences, 5 years in.

Why I Started

Apparently right now, ‘Newsletters are having a moment’. I’d love to claim credit for being ahead of the curve on this one, but those pieces were coming out in 2019, 2017, and 2015 as well. I can tell you that it didn’t feel ahead of the curve to start emailing people in 2016.

The newsletter grew out of teaching. I was one of the earlier lecturers at CUNY’s Master’s in Data Analytics (MSDA, later renamed MSDS, here are my students’ final projects, which I love and have proven surprisingly durable).

The students in the CUNY program had diverse experiences, the first class especially: some were just out of undergrad programs, and some were older professionals looking to make a career change, but the backgrounds were all over the place. One told me he had been stuck at Rikers Island for weeks before starting class, and another was a retail clerk, trying to break in to a more analytical role.

I loved the teaching and the students, but I was frustrated with the career placement support from the university, which they were trying to build basically from scratch.

At the same time, the job market for Data Scientists in 2014 and 2015 was wildly different from today (before the junior talent bubble). In NYC, tech companies were still paying much less than quantitative finance jobs and the sector was doubling every 3-4 years.

I was getting weekly inquiries from industry friends, looking for junior hires, and I was able to connect a few of them, but I couldn’t keep track of who was looking for a job and who would be a fit for which job. After a year of struggling through this situation, I decided just to offer everyone access to everything. I sent this email to all my students over the previous few semesters and about 25 people I knew in the industry:

19 people signed up for my first email.

By the 5th email, I was including links to events, though I didn’t make that permanent until July: if memory serves, the main reason I started was that I had missed PyData NYC in May and I couldn’t find a comprehensive list of data events in the area, so I extended the newsletter’s mandate to include events. I started including local posts a month later, since I’d written something, as had my friend Dan.

Since then, I’ve made a few minor changes, but the newsletter hasn’t changed much at all.

Pretty early on, people were telling me that they were finding jobs and candidates. Two of my students found jobs through the letter, including the one who had been at Rikers, and I felt a real rush. I was hooked.

Some Numbers

The questions I get asked most about the newsletter are ‘How much do you charge people to be included?’ ($0, though ODSC East offered to give me a pass this year before they went remote) and ‘How many subscribers do you have?’ (at the moment, 1,318). Given the topic of the email, obviously I have a bunch of other numbers that I track, mostly thanks to the undocumented tinyletter API & Jeremy Singer-Vine’s nifty library.

Starting off with some good old-fashioned vanity metrics:

207 emails written
122,723 emails sent (that's sum(email x list size))
71,483 unique opens (58%)
152,076 link clicks (2.1 link clicks per open)
175 lifetime unsubscribes (~ 0.84/letter)

The number of people signing up for the newsletter has grown in fits & starts, but has been more or less linear over time. I’ve spoken to a few other people about their letters (Jeremy makes all his data public) and unless you’re actively investing in growing a newsletter, I think this is a pretty common experience.

Open Rates for the letter had actually been more stable than you might expect: after an initial ‘burn-off’, open rates were basically stable for 2 years, from mid-2018 until earlier this year.

There actually has been a pretty large drop-off in late May / early June, where open rates fell pretty meaningfully. This has been something of a mystery to me. I’ve been able to rule out a few potential causes (getting caught in spam/filters, less interested new subscribers) but I still have a few suspects:

More competition: there are lots of data newsletters out there, (newsletters are having a moment, haven’t you heard?) I know I can’t keep up with everything that’s going on out there. It stands to reason that the newsletter would probably fall off peoples’ reading lists.

There’s some evidence for this: Data is Plural had a similar dip, though less pronounced. It also ‘makes sense’. However, I don’t think that explains the sudden-ness of the drop.

Less Value: it’s possible the newsletter became less valuable to people, through several different potential avenues.

One, there are no longer in-person events. Anecdotally, many people have told me that they find the event listings to be the most valuable part of the newsletter. This makes sense to me, but the timing doesn’t match up great: I stopped posting in-person events months before the open rate dropped,

Two, there’s the possibility that having a newsletter about NYC is less valuable to people because many of them are moving out of NYC. This is the most depressing possibility, in my mind.

Lastly, I can’t rule out that the difference has been me, and changes in the letter’s quality & focus. The dropoff happened right after the murder of George Floyd, and in the months after I made more references to legal & ethical considerations than before (though I certainly had made them before) and I can’t rule out that it may have pushed readers away.

It’s also possible that there’s been a drop in quality broadly, because of (gesturing vaguely at 2020) life. This was a tough year for lots of people, and though I’m certainly very fortunate, I can say that I’m no exception. I don’t think I did the best I could have, every single week.

That got a little heavy. Let’s look at the top links!

By raw clicks :

IBM will offer free COBOL training to address overloaded unemployment systems I totally forgot about the long waits to sign up for unemployment checks in March and April, and that systems were buckling under the load.
Our remote work future is going to suck More feel good stuff! In fairness, this is probably here mostly bc I messed up the link
A job posting for the ACLU Now we’re talkin’.
Working for the Vera Institute of Justice
Data Scientists Should Be More End-to-End a great piece by Eugene Yan

If I look at links by open rate, it’s mostly the same, though I’d also include this great post on Engineering Career Development at Etsy and my offhand link to Jewish Mysticism (my next newsletter? Who knows?!)

I also spent some time looking at what types of entities in the subject line lead to a greater open rate. I haven’t come to great conclusions despite a fair bit of work on it. It appears that subject lines that highlight legal, ethical, & privacy considerations do well, while those linking to insurance or finance jobs do poorly, which makes sense. It also seems like having subject lines that are a little off-topic don’t wildly impact open rates, but they do lead to more people unsubscribing. I’m still playing around with the data.

Observations & Advice

So that’s enough about my newsletter! What advice do I have for potential future newsletterers?

Do something you think is fun and important: This is very obvious advice. But it’s easy to focus on the numbers behind a newsletter, and there’s obvious value in ‘being big’. But in general most newsletters won’t make it onto a leaderboard or a ‘10-best’ list. Doing something that you enjoy and get meaning out of is critical to get you through the inevitable highs and lows.

To put maybe a finer point on it: writing regularly is a lot of work, no matter what. I have set up my newsletter to be the absolute least amount of work I can possibly make it, and some weeks I still find it to be too much. Some weeks no one responds and I feel like I’m just shouting into a void (that happens less now, with more subscribers). At Gawker, I could see the abandonment rates of Kinja blogs, and the numbers are staggering: 1-year survival rates were under 25%, even for the core audience.

Find a rhythm that works for you, don’t sweat deadlines: I know every audience is different, but for my audience, the timing of the email makes very little difference. Yes, email performs better on some days than others, but the effect is very small (except Fridays & weekends). I’ve played around with sending the letter at different days, and different times of day, and at different cadences, and the effects aren’t large. If I had to guess a ‘best time’, I’d say early evening letters have the highest engagement rates, but you’ll probably notice that I usually don’t send letters at that time because it doesn’t really work for my life.

I suspect the drop-off in clicks on weekend emails is due to more people opening on mobile devices, but I don’t have that data ¯\_(ツ)_/¯

Value the relationships: I read somewhere (I can’t find the source now) that the value of a lot of writing is not necessarily in the volume of people who read your work, but in the number and quality of conversations that it spurs. This has definitely been the case for me.

Email is different from social media in that emails are personal correspondences with a real expectation of privacy. The inbox is a safer, more private space (that unfortunately says more about social media than email). Email newsletters have a lot more friction and won’t ever achieve the broad visibility of a great tweet, but I’ve built many more important relationships via my 5 years with a newsletter than my 11 years on Twitter (despite spending much more time on the latter).

Closing Thoughts

One question I don’t get asked much, but I wish I did, was ‘How many people have found jobs through your letter?’

I don’t know the answer to that, exactly, but it’s in the dozens. It may be 100 at this point, though I doubt it. I know we’ve hired at least 3 people onto the Squarespace data team primarily because of the newsletter in less than 3 years, so in that regard it’s been a great success (want to be #4?).

I have a few thank-you notes that I’ve gotten from people who have found jobs & consulting gigs, or who have hired someone, through the newsletter. Those are gold. I find myself occasionally looking at them when I don’t feel like putting the letter together.

So, thank you to everyone who has sent me a note like that. And thank you to all of you who have sent me a posting, or an article, or an event to include in the newsletter. I literally couldn’t do this without you all. And to any subscribers, if you have ideas on how to make the letter better, please let me know, either on twitter or just reply to one of my emails.

Here’s to another 5 years!

(Thanks to Jeremy Singer-Vine for his comments on a draft of this post!)