Pickuma Newsletter Growth: Six Months of Subscriber Metrics, A/B Tests, and Lessons

I launched the Pickuma newsletter in January 2026 with exactly one subscriber: me, on a test email, to verify the deliverability pipeline. Six months later, the list is 2,400 subscribers with a 44.1% average open rate and a 12.3% average click rate. This article is the data behind those numbers — what worked, what did not, and the acquisition channels that produced 90% of the growth.

I had no social following, no existing audience, and no budget for paid acquisition when I started. Every subscriber came from content-driven channels, and the growth curve reflects that slow, compounding reality.

The Growth Curve: Zero to 2,400

Month one (January 2026) added 87 subscribers, almost entirely from the “subscribe” link at the bottom of the first batch of published articles. I did not promote anything. The articles had barely any traffic, so the conversion rate from readers to subscribers was a vanity metric built on a tiny denominator.

Month two added 140. Month three jumped to 310, driven by a spike in organic traffic after a GitHub Copilot review started ranking. The in-article CTA — a one-line text link reading “Get the next review in your inbox” embedded in the article body — converted at 3.8% of readers who scrolled past it. That same CTA, placed at the bottom of the article below the FAQ, converted at 1.2%. The placement mattered enormously.

Month four added 420. A Supabase article hit the front page of Hacker News for about four hours and drove 6,000 visitors in a day. The newsletter CTA on that page converted at 1.7% — lower than the average because HN readers are transient, not loyal. Of those 102 HN-driven subscribers, 18 unsubscribed within 30 days. The attrition rate on viral traffic subscribers was 17.6% versus 4.1% on organic search subscribers. People who find you through search stay. People who click through from aggregators are shopping.

Month five added 590. Multiple articles were now ranking on page one for long-tail developer queries, and the compounding effect of every article carrying a CTA started to show. Month six added 853 on pace, and the 2,400 total represents what I would call the end of the cold-start phase.

Acquisition Channels: What Built the List

I tracked subscriber origin with a simple UTM-based system: each CTA on the site uses a distinct UTM tag that maps to the placement and the article. Here is the channel breakdown as of May 2026:

In-article CTAs accounted for 61% of all subscribers. This includes both the mid-article text link and the bottom-of-article inline form. The mid-article link outperforms the bottom form by roughly 3 to 1 in conversion rate but appears in fewer articles because I only include it when the article is long enough for a natural break point.

Direct signups from the newsletter landing page accounted for 14%. This is the URL I share on social media and in guest appearances. It converts at 6.1% of visitors — high because anyone visiting a dedicated landing page has already made the decision to consider subscribing.

Cross-promotion from other developer newsletters accounted for 11%. I reached out to three newsletters in adjacent spaces — one covering dev tools broadly, one focusing on cloud infrastructure, and one on Rust — and offered a simple swap: I would recommend their newsletter in an issue of mine if they would do the same. Two of the three agreed. That produced about 265 subscribers over three months, and the quality of those subscribers (measured by 60-day retention) was comparable to organic search subscribers.

Social media (Twitter, Bluesky, Mastodon) accounted for 8%. I am not a prolific social poster — roughly twice a week I share a new article or an interesting finding. Social subscribers convert at the lowest rate and churn at the highest rate, similar to aggregator traffic.

The remaining 6% came from miscellaneous sources: GitHub README links, conference talk slides where I mentioned the site, and a handful of Reddit comment threads where someone else linked to an article and readers found the subscribe form.

Email Performance Metrics

I use Buttondown as the ESP, and the analytics are straightforward. The newsletter goes out every two weeks — I experimented with weekly, but the open rate dropped from 44% to 31% at weekly cadence, and the unsubscribe rate doubled.

The biweekly cadence settled at these averages across the last 3 months: 44.1% open rate, 12.3% click rate, 0.6% unsubscribe rate per send. The list is small enough that individual articles swing these numbers, but the pattern is stable.

The click rate breaks down interestingly by link type. Links to new reviews and comparison articles get 8-9% click-through. Links to meta articles like this one get about 5% — not surprising, since meta content appeals to a subset. The highest-performing link type is “what we learned” or “lessons from” content, which averages 14% click-through. Readers click on conclusions more than announcements.

A/B Test Results on Subject Lines

I ran eight subject-line A/B tests with samples of 200 recipients each. The tests compared three formats: descriptive (“New Review: Supabase vs Firebase 2026”), curiosity (“The database decision most teams get wrong”), and numbers-driven (“2,400 subscribers, 44% open rate, 1 lesson”).

Descriptive subject lines won five of the eight tests by a margin of 12-18% in open rate. The curiosity format won two tests — both for articles about controversial topics where the curiosity was genuinely earned. Numbers-driven won once, for the six-month retrospective article where the metrics themselves were the story.

The headline: developer audiences are trained to distrust clickbait. A subject line that tells them exactly what the email contains performs better than one that tries to entice them into opening it. This aligns with how developers evaluate any information source — specificity signals competence, vagueness signals marketing.

I stopped A/B testing after eight rounds because the pattern was clear and consistent. The newsletter now uses descriptive subject lines for every issue, with format “[Pickuma] New Review: [Tool Name]” for review sends and “[Pickuma] What We Learned: [Topic]” for meta sends.

What Did Not Work

Lead magnets. I created a “Best Developer Tools 2026” PDF checklist and offered it as a download incentive for subscribing. It added 28 subscribers in three months and took six hours to produce. The conversion rate on the landing page with the lead magnet was 5.9% — lower than the no-lead-magnet page at 6.1%. Developer audiences do not want another PDF. They want the next article.

Exit-intent popups. I ran one for two weeks. It generated 14 subscribers and a handful of complaints. The signal-to-noise ratio on exit popups is bad for any audience, but particularly bad for developers who read technical content and have a low tolerance for interruptions. I killed it.

Welcome sequences. I built a three-email onboarding sequence with links to the most popular articles. Open rates on the sequence were 62%, 41%, and 28% across the three emails — strong decay, which is normal — but the click-through rates were 3% or below for all three. Subscribers who joined through a specific article already knew what they wanted. A generic welcome sequence added nothing. I replaced it with a single welcome email that says “Here is the archive, here is the RSS feed, reply if you need anything.”

Segmentation and Personalization

At 2,400 subscribers, the list is too small for meaningful behavioral segmentation. I tested a simple split: subscribers who clicked a review link in the past 60 days versus subscribers who had not. The engaged segment had a 57% open rate and an 18% click rate. The disengaged segment had a 21% open rate and a 4% click rate. The test confirmed what every email marketer already knows — engagement is bimodal — but quantifying the gap was useful because it told me when to prune.

I implemented a 90-day re-engagement rule at month five. Subscribers who have not opened any of the last six issues receive a one-time re-engagement email with the subject line “Still interested in developer tool reviews?” and a single-click option to stay or unsubscribe. About 14% of recipients click “stay” and immediately open the next issue. About 23% unsubscribe. The remaining 63% do nothing, and I remove them from the list 30 days after the re-engagement email.

This pruning reduced the subscriber count by about 8% but increased the average open rate from 41% to 44%. The raw subscriber number is a vanity metric. The open rate and click rate are the numbers that determine whether the newsletter is a useful distribution channel, and a smaller, more engaged list is a better channel than a larger, disengaged one.

I do not segment by tool interest or category because the subscriber base is too small for the segments to be statistically meaningful. When the list crosses 5,000, I plan to add a preferences page where subscribers can opt into specific categories — AI dev tools, infrastructure, SaaS productivity, and meta — and I will measure whether category-specific sends outperform the general newsletter. Until then, every subscriber gets every issue.

What I Would Do Differently

The single biggest mistake in the newsletter’s first six months was launching before the site had enough content for new subscribers to browse. The first 50 subscribers joined a newsletter that pointed to a site with 8 articles. Their first experience was reading a newsletter, clicking through to the site, consuming the backlog in 20 minutes, and then waiting two weeks for the next issue. The unsubscribe rate on those early subscribers was 31% in the first 60 days.

If I were starting over, I would publish 15 articles before collecting a single email address. The newsletter is a distribution channel for content, and a distribution channel with nothing to distribute is a churn machine.

The second mistake was not collecting subscriber source data from day one. I added UTM tracking in month three, which means I have no attribution data for the first 300 subscribers. I know they came from the site because there was no other acquisition channel, but I do not know which articles converted them, which CTAs they clicked, or how their retention compared to later cohorts. The data I am missing from months one through two would answer the most important growth question: which article type produces the highest-quality subscriber.