An Announcement, a 2.5 Year Newsletter Update, and My Highest Ever ROI Action at Amazon
My move to Substack, a bit about how the newsletter has grown, and a great story about how I helped drive down errors across dozens of systems, without my team lifting a finger.
Hello there, I’m Dave Anderson! Welcome new folk, and long-time readers! This is my weekly newsletter where I share stories and advice for both technology veterans and newcomers on how to grow their careers, build their leadership capabilities, and level up their interviewing skills.
A brief introduction first.
I joined Amazon in 2007 as a Level 5, the lowest level engineering management position. By the time I left Amazon in 2020 (to join Jeff B’s early education nonprofit as CTO), I had been promoted to Technical Director & General Manager.
I’d managed hundreds of engineers, interviewed over a thousand candidates (hiring hundreds), wrote promotion documents for dozens of engineers, and participated in many more dozens of promotion review meetings. I was a member of the Bar Raiser Core organization, which means I helped determine how the bar raiser program would be run, and how it would evolve.
Writing a book has been a dream since I was a child. When I left my CTO position, I decided that writing would be my next adventure. To encourage myself to write consistently, I started a newsletter. The topics would be varied (probably a bit too much), but mostly things I learned while working at big tech (mostly Amazon) while growing my career, and the careers of others.
I didn’t expect the wonderful reception from readers across the world. I found that newsletters are a highly scalable way to share information. Having worked at Amazon, I’m addicted to mechanisms that scale.
This all briefly explains my history, and hopefully helps establish some credibility regarding my favorite topics.
I love career development, particularly how to be personally successful in technology organizations.
I believe interviewing is a learnable skill; one you can practice and perfect.
I have always insisted that every employee (not only managers) should demonstrate leadership. The skills and behaviors necessary to lead others is something that everyone should learn.
I’m confident that most managers receive zero formal education in management. And I love helping people take control over their career development.
I have three major things to go over today.
A Newsletter update.
A story from Dave. What I did to achieve the best results (considering the effort invested) during my time at Amazon.
Here we go!
The Announcement — A Move to Substack
When I began the Scarlet Ink Newsletter (see above, after leaving my 9-5 job), I started with the Ghost platform. I liked the fact that it was open source, and that I could completely own the platform (code and all).
Over the last few years, Ghost served me fine. I used some of my spare time on LinkedIn and Twitter to communicate with potential readers (you folk!). Between those social media channels and subscribers recommendations, I had continual growth over the last ~2.5 years.
When I originally picked Ghost, I had made a spreadsheet of Ghost vs. Substack. I listed the features I wanted, and the projected costs of using either platform. I eventually decided that Ghost had enough of the features I desired.
However, a couple of major things happened over the last 2.5 years. And those changes have made me change my mind.
Substack Features — When I listed those features in my spreadsheet, I knew that Substack already had more features. What I didn’t think about was that Substack had funding and significantly more revenue. This means that Substack’s feature list grew at a faster rate than Ghost. I’ve repeatedly seen Substack features launched where I thought, “Oh, that’d be nice! (Sad face)”
What particular features seem useful right now? Here are a few:
Search. I know my long-time readers will be thrilled to see this. I’ve gotten so many emails from you all, asking me why I don’t have a search feature! Unfortunately, Ghost doesn’t have full-text search out of the box. This is a big gap I’m excited to have solved.
Support. When my readers on Ghost have issues, they contact me. When readers on Substack have issues, they can contact Substack! What a win for readers and writers!
Group purchases. I regularly have people ask if they can purchase group subscriptions for their team. I can say yes going forward.
Gift subscriptions. Same thing! I know a number of you have asked to be able to give away subscriptions. I’ve manually setup gift subscriptions for people who asked really nicely on Ghost, but it was a lot of manual work. I love that it’s built into the platform. See? Look at this button!
Substack Ecosystem — One of Ghost’s benefits is also the biggest flaw. On Ghost, I was an island. I liked my independence. I had my newsletter, and others had their newsletters. My readers would find out about another newsletter if I purposefully linked to that person’s newsletter. But my writing→ email → customer is not a platform, it’s a communication mechanism. This means that regardless of how many customers use Ghost newsletters, I don’t benefit at all. All growth is directly by my action, for better or for worse.
On the other hand, Substack has great recommendation features. It can say, “Hey, if you liked Dave’s writing, you might enjoy…” And as long as they continue to build it carefully (and avoid spam), it has become a magnificent ecosystem of writers recommending writers, and readers recommending writers. This has only grown over the years as Substack newsletters continue to grow, and the Substack mobile app has become a big success. And I wanted to be a part of that. Because if I can spend more time writing articles and less time focused on social media marketing, it’s a considerable win in my book. And it’s a win for my readers as well. More / better content rather than me spending mental energy on selling.
Part of that Ecosystem includes ways that Substack readers can interact with articles. You can comment, you can make Substack notes. Previously, the Scarlet Ink community was whoever responds to my LinkedIn or Twitter posts. Soon there will be better ways for readers to interact! I’m excited to see how that pans out.
The move to Substack happened the evening of 12/4.
Fingers crossed, the site and all the archived newsletter articles appear to be on Substack as of right now. I found a few broken links and fixed them.
This update shouldn’t require any work on your part. If you do run into a problem or a question or a concern or have a suggestion, please reply to my newsletter, and I’ll excitedly figure out how to help you. Because I enjoy receiving emails.
A Newsletter Update
I wrote one of these around 1 year ago, and I figured this was a great time to write another. I’ll keep it short, as otherwise I’m going to make this article far too long.
I started the newsletter on May 10th, 2021, and had 6 (free) newsletter subscribers on day one.
1.5 years in, I had 23k free subscribers, and almost 900 paid members.
2.5 years in, I have 26k free subscribers, and over 2.4k paid members.
Why did the paid list grow nicely, but the free subscribers list didn’t grow much? I’ll explain that a bit later.
Great results from social connections
Gergely offered to post one of my articles on The Pragmatic Engineer as a guest post, and I ended up using his state of the tech market newsletter as one of my articles.
At the same time, Alex from ByteByteGo offered to post one of my articles on his newsletter as a guest post. Which was a massive coincidence, but awesome timing!
I'm extremely grateful to Gergely & Alex for their friendly contributions to my newsletter. With that combined attention, I saw a substantial spike in subscription traffic for the following two weeks. It’s pretty clear that making friends and connecting newsletters together creates a wonderful network effect. See the Substack Ecosystem section above.
Personal referrals are a large source of subscriptions, I suspect
As my newsletter has grown, so has the rate of new paid subscribers. At the same time, my social media posts aren’t significantly more popular. What could be causing the continual increase in new subscriptions? It suggests that having paid subscribers drives new paid subscriptions.
I've heard anecdotally from many of you paid subscribers that you send my newsletters occasionally to co-workers or friends. I suspect that is the key to my continued growth.
So thank you so much to those of you who have sent links to my articles to other people!
Free Subscriber Mistakes
Since the end of 2021, I’ve only had a paid newsletter. Free members could still read the first ~50% of my articles.
However, I have been overly frugal at times, to my detriment. Ghost’s model charges per member (including free members). As a way to keep my costs down (Amazonian’s might call it a frupid decision - frugal to the point of stupidity), I repeatedly purged my free list of free subscribers who (per analytics) weren’t opening emails.
I figured I could stop emailing them, and save myself money. Now, I’m not entirely sure if those analytics were accurate (as analytics are blocked by multiple platforms), but I ended up deleting many thousands of free subscribers. Ooops.
I stopped the purge process. As with many other things, I’m going to let Substack handle account management. Yet another benefit of moving to a new platform.
What I did to achieve the best results (considering the effort invested) during my time at Amazon.
My initial two years at Amazon were spent in the global payments group. It was a platform team deep inside Amazon’s technology stack. As a side note, platform teams are a great way to get to know a company. I learned a lot about how Amazon works through my time in global payments. We interacted (at least lightly) with any organization at Amazon that wished to charge money. This was essentially every organization.
After those two years, I moved to the Marketplace group, where I was placed in charge of Seller Central as the line manager (direct manager of engineers). Seller Central is the website that 3rd party sellers use to manage their businesses on Amazon. Per Amazon, over 60% of Amazon’s sales are from 3rd parties. Anyway, my point is that it’s a big deal.
What exactly did I own?
While I ran the site Seller Central, I didn’t run every function. There were teams for order management, inventory management, customer service, and shipping. All in, there were at least 40 teams directly tossing features into the Seller Central website. And I had to own that thing. Not really any of the features, but the website itself.
Now, in Amazon terms, what did owning Seller Central mean? Well, in the collective opinion of my manager and other senior leaders, it meant that Seller Central needed to be a great tool. It needed to work properly, fast, and consistently. Even if the tools and systems integrated into my website were owned by dozens of my peers.
What was business as usual?
When I joined the team, we had a general plan of action. My team had systems monitoring the error rates on our website. If the errors spiked too high, we’d try to figure out what caused the error spike. Then we’d report it to our leadership team. For example:
“Two customers in NA ran automated jobs to load their entire order history. This created an unexpected load on the order archive system. Due to the volume and speed of customer requests, the order archive system had increased latency for the 47-second duration of the event.”
I’d then link to the trouble ticket for the order archive system. That email was sent to the S-Team, among others. Whew! Now that’s visibility for you.
Except that no one knew who owned the order archive system. Any questions from the S-Team would come to me. And I’d find out the answers from the corresponding team, write them up nicely (because that’s one of my things), and write back to the S-Team.
Engineer: “Shit went south. I mean seriously, this system is a pile of unmaintained crap.”
Dave: “The order archive system had an outage.”
I got some nice visibility, but it didn’t solve things.
How do I improve things?
I remember a conversation going something like this.
My manager: “Hey Dave. I see the error rates on Seller Central are looking pretty bad these days. I think you should take an S-Team goal to improve them.”
For context, an S-Team goal is essentially a goal where you need to report on its status up to the S-Team. You only want these goals if you’re guaranteed to achieve them because otherwise they’re a huge pain in your tushy.
Dave: “That sounds great! Except I don’t control everyone’s error rates, I only report on them. So I probably shouldn’t take that goal.”
Manager: “That sounds like you have a great challenge ahead of you! What percentage will you improve those error rates?”
Dave: “Well shoot. They’re increasing these days, not decreasing. How about 10%?”
Manager: “10% is boring. This is an S-Team goal! You need to be bold. How about 30%?”
Dave: “Well, I feel that sounds super hard and not at all good for my career.”
Manager: “I have confidence in you. Thanks!”
So I ended up with a goal to improve the error rates by a significant margin, but no clear way to do it. I didn’t own the pages, or the systems behind the pages. My manager didn’t own them either. Or my manager’s manager. We’re talking about different S-Team members for some of these systems.
However, I agreed with my manager’s sentiment. Someone needed to figure out how to create a better experience for Sellers. Each individual team wasn’t incentivized to resolve recurring problems, as their occasional errors were low on their priority list. What we needed was a collective solution. And I was in the most reasonable leadership position to find that collective solution.
Introducing the Monthly Error Report.
What you measure, you improve. It’s a rule I deeply internalized during my time at Amazon.
I thought carefully about the error rates. The issue as I saw it was an issue of visibility.
Teams couldn’t see their error rates over time. They’d know about big events, but not “how many errors did I generate this month per million customer requests?”
Senior Leaders of teams (Directors and VPs) couldn’t see if their various teams had robust systems, or if they needed operational investments.
The S-Team didn’t have any visibility into the availability of our seller systems beyond specific events.
Sellers would complain to our Technical Account Managers. That complaint would fall into a black hole because those TAM employees had no data.
I recognized that I needed to connect error rates → systems → leaders. And then give visibility to the whole thing.
I created a simple data table in an email. In that table, it listed the top contributors to our collective error rate in each region (ordered by error rate). It listed the page / function name (e.g., order archive system). It listed the organization’s name that owned the page / function, and most importantly, the name of the Director or VP in charge of that organization.
I warned all the organizations that I was going to start reporting in this way, and created my first monthly report. It was sent to the S-Team.
I still remember a Director rushing into my office. They quickly asked, “How much time do we have to fix this before the next report goes out?!”
There was an immediate and obvious drive across a dozen groups to get errors fixed. Teams fought to avoid being the top error contributor for the month. Our error rate plummeted week after week. All with zero engineering effort on my team. I was literally just sending an email, which I had built with Outlook and Excel.
A few months into the program, one of my monthly error reports was sent out, highlighting the plummeting error rates. It got a smiley face from Jeff B. Which is about as much praise as you can ever expect from Jeff. So that was super cool.
What did I learn from the whole thing? Absolutely everyone involved was a good person. They wanted to build features, improve their systems, and have good operations. There’s just not enough time for everything. And when you begin to cut things out of your schedule, you always cut the things which are not measured. You fix the squeaky wheel, as the saying goes.
The corollary is also true. The reason this drifted for as long as it did was due to inadequate measuring, reporting, and communication. Our sellers were annoyed, but Amazon teams didn’t prioritize those issues because my team wasn’t making those issues visible enough.
While you can’t abuse it, becoming the squeaky wheel deliberately is absolutely a viable strategy. I put a bullseye on the leaders of a few organizations, and they immediately did what you would expect. They pushed their teams to remove that bullseye. As one team’s manager said, “My Director got an email from his VP, asking why we were at the top of the list. So I need to fix this right now.”
I hope you have a great day!
If you experience any issues accessing your account, receiving emails, viewing the website, or perhaps you’ve lost your dog, please contact Substack support! How exciting to have a support team!
For clarifications about anything from me, such as the Substack move, something from one of my articles, etc, please feel free to reply to any newsletter, and I’ll be happy to respond!