February 14, 2025. My account averaged 545 impressions per day. I gained 0.6 followers daily. More than half my days showed zero or negative growth. I had no strategy, no tracking, no system. I posted AI images when I felt like it and hoped the algorithm would notice.

February 14, 2026. My account averaged 10,095 impressions per day. I gained 18.9 followers daily. 97.2% of my active days showed positive growth. I'd secured the Adobe Firefly Ambassador program, published an article that earned 462 bookmarks, and built a community of AI creators who actually engage with each other.

The transformation between those two dates wasn't a lucky break. It was eight distinct phases, each with its own lesson. I tracked every day. Here's the full story.

Both measurements are 14-day windows from the same account. Same niche. Same tools. Same person making weird photorealistic surrealist AI art. The only thing that changed was the system behind it.

The bookmark growth is the number that tells the real story. 8 to 690. An 85x increase. Impressions grew 18.5x. Follows grew 29.3x. But bookmarks grew 85x because the content shifted from "look at this cool image" to "here's how I made this and how you can too." The audience stopped just admiring and started saving.

If you're an AI creator at the beginning of this curve, everything in this article happened in 365 days. Not five years. Not with a team. One person, documenting the process of making AI art, building community around it, and tracking what worked.

252 days. +0.9/day. 1,290 avg daily impressions.

This is the phase most AI creators live in permanently. Posting sporadically. No engagement strategy. No tracking. No idea why some posts perform and others don't.

54% of my days in the dormant era showed zero or negative growth. 49 days were net negative, meaning I lost more followers than I gained. The account was slowly eroding.

On August 28, I posted a single image that hit 35,561 impressions. My biggest day of the entire dormant era. It generated +23 net followers. By September, I was back to 0.5/day.

Viral moments during the dormant era were fireworks: spectacular, temporary, and leaving nothing behind. Without a system to capture the attention, 35,561 impressions produced a day of growth and eight months of nothing.

On July 15, I lost 35 followers in a single day (38 unfollows, 3 new). The spam flag. The "reply guy" label. Creator revenue paused. That was rock bottom. The account was running on inertia and the inertia was negative.

12 days. +10.2/day. 5,452 avg daily impressions.

On October 22, I set a goal: 5,000 followers to qualify for an Adobe Firefly Ambassador referral. At the time, I had roughly 2,170 followers. The goal seemed absurd.

The first thing I did was simply show up. Consistent daily posting. Daily engagement with other AI creators. Nothing strategic yet, just presence.

The results were immediate. Daily impressions jumped from 1,290 to 5,452. Growth went from 0.9/day to 10.2/day. That's an 11x increase in follower growth from nothing more than consistency.

For AI creators: this phase proves that the bar for "better than dormant" is shockingly low. You don't need a strategy to start. You need daily presence. Post your work, engage with the community, repeat. Strategy comes later. Momentum comes first.

5 days. +11.6/day. 9,784 avg daily impressions.

This is the most important phase of the entire year. Not because the growth was best (it wasn't) but because this is where I started measuring.

November 10 was the trigger day. My account generated 9,930 impressions but only 26 profile visits and 4 net followers. I calculated the curiosity rate: 26 visits divided by 9,930 impressions = 0.26%.

The industry benchmark for curiosity rate is 1 to 5%. I was converting at the bottom 5% of the range. My content was good enough to get impressions. My profile was converting decently (6 follows from 26 visits = 23%). But almost nobody was clicking through to see who I was.

The math of potential lost: at 0.26% curiosity, I got 26 visits and 6 follows. At 1% curiosity, I'd have gotten 99 visits and 23 follows. At 3% curiosity, 298 visits and 69 follows.

I was leaving 17 to 63 followers on the table every single day. And I never would have known without the data.

For AI creators: you might be in this exact situation right now. Good content, decent engagement, but invisible to new audiences because nobody is curious enough to visit your profile. The fix isn't better art. It's better framing.

10 days. +24.6/day. 8,094 avg daily impressions.

Armed with the diagnostic, I made one change: I replaced every generic call-to-action ("Like, share, bookmark and follow for more") with specific value propositions. "Technique #3 of 47 I've documented. Full library on profile."

The 5-day result: +97 net follows versus +39 from the previous five days. A 149% increase from changing the CTA language alone.

Then I discovered the three-part growth formula:

When all three were present on the same day, growth averaged +31 to +41. When only one or two were present, growth averaged +12. The formula was consistent and repeatable.

This was the best 10-day stretch of my entire year. +246 total followers. For an AI creator at 2,200 followers, that's transformative. And it came from systematic testing, not from any single viral post.

28 days. +19.3/day. 10,876 avg daily impressions.

This phase proved the formula wasn't a fluke. Four straight weeks of consistent execution. Hit 2,500 (November 21), 2,750 (December 3), and 3,000 (December 18).

The shoutout discovery emerged here: 1.08 follows per shoutout post, 3x better than any other content category. Regular shoutouts to other AI creators built reciprocal relationships, expanded niche reach, and established a reputation as someone who curated quality work.

For AI creators: Phase 5 is where most people want to start. But the diagnostic (Phase 3) and formula (Phase 4) are what made this phase possible.

14 days. +5.3/day. 3,931 avg daily impressions.

I went from +20.0/day (December 15-18) to +5.3/day over the holiday period. Growth dropped 73% in a week.

Some of this was platform-wide. X activity drops significantly during the holidays. But I also took breaks. Skipped days. Reduced my reply volume. Let the 50+ daily conversations drop to maybe 10.

The compound effect was immediate and brutal. Impressions fell from 10,876/day to 3,931. Bookmarks fell from 21.3/day to 5.1. The algorithm saw reduced activity and reduced distribution accordingly.

December 25: zero net growth. December 29: zero net growth. These were the only zero days in my entire 107-day active period. Both during the holiday crash.

For AI creators: breaks are necessary. But understand what they cost. Missing 2 to 3 days creates a momentum penalty that takes 1 to 2 weeks to recover from. If you're going to take time off, plan for the recovery period. Don't assume you'll pick up where you left off.

30 days. +11.0/day. 5,595 avg daily impressions.

This phase had two sub-phases that tell very different stories.

January 2 to 13 (Recovery): +15.3/day. I came back strong. Resumed the formula. Posted consistently. Growth recovered quickly, nearly reaching pre-holiday levels.

January 14 to 31 (Algorithm Trap): +5.2/day. Then I made a mistake. I shifted my content from AI art documentation to generic algorithm education threads. The threads performed well on paper: impressions were up (4,639 to 6,410 weekly average), engagement was solid. But growth collapsed.

Why? The algorithm education content attracted growth-hackers and social media marketers instead of AI artists. I published 31 algorithm posts during the trap period. They generated 14,299 total impressions but only 6 follows. That's 0.42 follows per 1,000 impressions, while my AI art content had been converting at 19.1 follows per 1,000 impressions on my best days.

The content was good. The audience was wrong. My impressions went up but my followers went down. Weekly growth declined three straight weeks: 14.9/day to 8.6/day to 5.0/day.

For AI creators: this is the trap that kills growth for accounts in our niche. The temptation to post generic "how to beat the algorithm" content is strong because it gets engagement. But the people engaging aren't your people. They won't follow an AI art account. They're following for algorithm tips, and when you go back to posting AI art, they unfollow.

14 days. +18.9/day. 10,095 avg daily impressions.

I snapped back to AI-art-anchored content. Published my comprehensive storybook workflow article on February 3. On February 4, Adobe accepted me into the Firefly Ambassador program.

The five-day viral window (February 3-7) produced +186 followers, 80,001 impressions, and 612 bookmarks. The article alone earned 462 bookmarks, generating 32.4% of all bookmarks I'd earned across 1,789 posts in the entire year.

February 4 set all-time records: 107 new followers, 32,978 impressions, 459 bookmarks, 3,129 engagements. Every metric at its highest point in 365 days.

The recovery from the Algorithm Trap was immediate. The week of January 26 to February 1 averaged 5.0/day. The week of February 2-8 averaged 27.6/day. 5.5x growth by returning to the content my actual audience wanted.

Here's the number that doesn't get enough attention: 97.2%.

Out of 107 active days (October 31 to February 14), 104 showed positive growth. 3 showed zero growth. Zero days showed negative growth. Not one.

Compare that to the dormant era: 54% of days were zero or negative. 49 days were net negative. The account was bleeding followers more days than it grew.

The difference isn't talent. I was making the same kind of AI art in both periods. The difference is system. Daily posting, daily engagement, daily tracking, daily community building. The consistency compound works in both directions: consistently absent produces decay, consistently present produces growth.

For AI creators who feel like they're not growing: are you actually present daily? Not just posting, but engaging. 50+ replies. Shoutouts. Community conversations. If you're posting images and logging off, you're in the dormant era whether you realize it or not.

Across the full 365 days:

1.19 million impressions. Nearly 2,000 net followers gained. Over 2,000 bookmarks. And 80,041 individual engagements with my content across a year.

But those numbers are misleading if you don't understand the distribution. The dormant era (252 days) produced +236 followers. The active period (107 days) produced +1,565. That means 42% of the year generated 84% of the growth. The other 58% was background noise.

The active period wasn't 107 days of perfection. It included a holiday crash (-73% growth), an algorithm trap (-66% growth), and multiple recovery periods. The 97.2% positive rate didn't mean every day was great. It meant the system was resilient enough to produce positive results even on bad days.

The Firefly Ambassador program was the goal that started everything. Adobe lowered the follower requirement, making it achievable at roughly 3,500 rather than the original 5,000. But the portfolio was what mattered: 40+ educational pieces, community building across six curated X lists, documented tool mastery, and the 462-bookmark article proving deep process documentation.

The 5,000 goal remains as a growth milestone. The Ambassador status was earned by the body of work, not the follower count.

If I could go back to February 2025, here's the sequence:

Week 1-2: Show up daily. Post your AI art. Reply to 50+ people in your niche. Don't optimize anything yet. This alone will 11x your growth versus sporadic posting.

Week 3: Start measuring. Track daily impressions, follows, unfollows, bookmarks, and profile visits. Calculate your curiosity rate (profile visits divided by impressions). Find your bottleneck.

Week 4: Fix the bottleneck. Replace generic CTAs with specific value propositions. "Technique #3 of 47" beats "Follow for more" every time.

Month 2-3: Build community. Shoutouts (1.08 follows/post). Curated lists. Reciprocal relationships. Become the person people follow to find the rest of the community.

Month 3-4: Document your process. Pick your best technique and write a comprehensive article. Every prompt, every setting, every failure. This single piece of documentation will outperform months of daily image posts.

Ongoing: Track bookmarks, not likes. Bookmarks correlate with growth at r=0.698. Likes at r=0.504. Create content people save.

I build guitars for a living. Woodworking is a craft where every joint, every curve, every finish layer represents accumulated knowledge passed down through documentation and demonstration. Nobody learns lutherie from seeing a photograph of a finished guitar. They learn it from watching the process, studying the joinery, understanding why the neck angle matters.

AI creation is the same craft with different materials. The finished image is beautiful. But the prompt sequence, the tool settings, the iteration history, the failures and adjustments: that's the knowledge that builds a following. That's what people bookmark, reference, and return to.

In a year, I went from 0.9 followers per day to 18.9. The art didn't change dramatically. The documentation did. The community building did. The systematic tracking did.

The tools exist. The niche is growing. The data is clear about what works. The only variable is whether you commit to the system.

0.9 to 18.9. 365 days. The journey continues.

Glenn is an Adobe Firefly Ambassador and AI creator documenting the craft of prompt engineering and creative process at @GlennHasABeard. He publishes The Render newsletter and creates the Stor-AI Time series adapting world folktales through AI-generated video.

This is the final article in a series analyzing one full year of X analytics: 365 days of account data and 1,789 individual post records. Every number is from official X analytics exports.