Social Growth
TheSequence Organic Social Growth
Growing a technical AI/ML newsletter's X account from under 1,000 to 46,000 followers without paid promotion.
Context
When I joined TheSequence in January 2021, it was a 6-month old AI/ML newsletter. It published dense, research-led writing for ML engineers and researchers, but it did not yet have a meaningful social media presence.
I was hired to grow Twitter/X first. As the channel developed, the work expanded into LinkedIn and adjacent growth tasks: improving how Substack content was organized, supporting outreach, and looking for ways to convert social attention into newsletter readers.
The audience was narrow and technical: AI/ML specialists around the world, including people at organizations such as Intel, Microsoft, Google AI, LinkedIn Research, Salesforce, Nokia Bell Labs, Samsung, MIT, and Stanford.
Challenge
The challenge was twofold: to build TheSequence’s social presence on X (Twitter) while attracting its target audience: practicioners, researchers and aspiring professionals in AI/ML field.
X (Twitter) was selected as a main platform because that target segment was actively using the platform. So we had to first understand the platform algorithm and posting rules and understand the target audience’s needs, language they speak and content they care about.
We had to start from scratch and develop the social media content strategy that was evolving as we were going deeper into the topic and collecting more information about audience response.
We worked completely organically with no budget for ads or promotion.
Role
I was initially responsible for building TheSequence’s presence on X/Twitter. As the channel grew, my role expanded to LinkedIn and the wider social-to-newsletter growth system.
I owned the work end to end:
- Developed the channel strategy and content plan, then handled daily creation, formatting, publishing, and scheduling.
- Created and validated flagship recurring formats, including research roundups, hiring digests, and curated course, book, and resource collections.
- Built a repeatable workflow for turning each newsletter issue into threads, standalone posts, and shorter updates that directed readers back to the newsletter.
- Tracked social media analytics, including metrics like impressions, profile visits, link clicks, follower growth, and other conversion signals.
- Used weekly cycles to refine the content mix and adapt topics, language, visuals, and post structure based on performance.
- Monitored AI/ML news and research, responding to timely industry discussions.
- Organized outreach, cross-promotions, and collaborations with AI/ML communities, experts, and relevant industry accounts to extend organic distribution.
This was a part-time role in a small team with two founders. I had substantial autonomy over the channel, with weekly reviews with my manager, Ksenia, to align priorities and evaluate performance. The work was fully organic, with no budget for ads or boosted posts.
Strategy
The strategy was built around four connected systems.
1. Newsletter-to-social repurposing
I created a repeatable workflow that turned each newsletter edition into threads, standalone posts, and short updates. This extended the value of research already produced for the newsletter without requiring new source material for every post, while connecting social content directly to newsletter acquisition.
The posts were designed to work as useful content on their own and directed readers to TheSequence newsletter content if they want to get more information ar dive into details.
2. Weekly data-driven optimization
Every week, I reviewed top posts by the key metrics. The key metric was profile visits, but I also combined it with impressions, bookmarks and link clicks. I doubled down on topics and formats that drove profile clicks that was a proxy for new followers or amplified reach, cut approaches that did not, and refined hooks, language, visuals, and post structure.
3. Flagship recurring formats
I built dependable series that audiences could recognize and return to: AI research roundups, hiring digests, weekly updates, and curated collections of courses, books, and technical resources. These recurring formats consistently drove the strongest profile-visit conversion.
The mix combined current research with evergreen educational resources. Technical information was made more scannable through clear structure, concise framing, and deliberate image and design choices.
4. Timely commentary and earned amplification
X/Twitter is shaped by fast-moving news and active industry conversations, so the workflow also had to respond to what the AI/ML community was discussing. I monitored research releases, company updates, expert talks, and posts from key figures, then incorporated relevant developments into the publishing plan.
Timely commentary and well-structured summaries helped attract reactions and reshares from the people and organizations being discussed. Combined with direct outreach, community collaborations, giveaways, and cross-promotions, this extended distribution beyond the existing audience and earned organic amplification from major accounts in the field.
Examples of Working Formats
The strongest recurring formats included:
AI updates of the week
Free books, courses, and technical resources
AI/ML hiring digests
ML research listicles
Threads with unexpected or insightful AI/ML facts and stories
Trending ML research and application reviews
Summaries of expert talks and updates, which sometimes drove reactions, reposts, and collaboration opportunities from the people featured
These formats helped the account become more than a promotional channel. It became a distribution layer for the newsletter and a useful feed for people following AI/ML research and industry movement.
AI updates of the week
The recurring roundup made current research and industry movement compact and easy to scan.
Free books, courses, and resources
Practical curation proved highly repeatable because each post offered immediate value to learners and practitioners.
ML research listicles
Compact paper collections made dense research easier to discover, compare, save, and share.
Threads with unexpected/insightful facts and stories
Trending ML research/applications review
Breakout resource posts
The largest single breakout featured a free deep-learning course and reached roughly 569K impressions, showing how far the right learning resource could travel beyond the existing follower base. A year-end recap of the most popular courses reached roughly 161K impressions, turning a seasonal collection into another significant wave of discovery and profile visits.
Results
The chart uses reported milestone values; the under-1K starting point is plotted at 1K for scale.
Audience growth
Follower count grew from <1,000 to 46,000. The curve was cumulative rather than dependent on one viral spike: weekly gains typically ranged from roughly 600 to 1,300 followers, reaching more than 3,900 new followers in the strongest month.
Reach and conversion
The account generated 20M+ impressions across the work. Monthly reach scaled about 60x, from roughly 37,000 to a peak of 2.38M impressions.
Top-performing posts reached 130K to 569K impressions and drove roughly 1,500 to 6,200 link clicks each, helping turn social reach into newsletter traffic and subscribers. The largest breakout featured a free deep-learning course and reached roughly 569K impressions. A year-end collection of popular courses reached about 161K, showing that the resource format worked both as an evergreen series and around seasonal moments.
Earned authority
The account also started earning organic reshares and engagement from people and organizations the audience already trusted, including Amazon Science, Ai2 (Allen Institute for AI), Hugging Face, Stanford University AI Lab, Harvard University, Gary Marcus, and sktime. Earned mentions grew to about 285 per month, building authority without relying on paid promotion.