I had a 6% engagement rate. Industry average is 1–3%. My content was getting bookmarked, liked, and replied to at rates most creators would kill for.

And I grew by 3 followers that day.

Not 3 hundred. Not 3 thousand. Three.

November 10, 2025. I'd been pushing content on X for weeks. AI storybook scenes built in Adobe Firefly, surrealist photorealistic images that people genuinely seemed to love. The kind of weird-but-beautiful stuff where a giant octopus is casually reading a newspaper in a laundromat and it looks like a photograph. People engaged. They bookmarked. They replied with fire emojis and "how did you make this?"

But almost nobody followed.

I was generating 9,930 impressions that day. Decent reach. 380 engagements. 18 bookmarks, proof that people found the content worth saving. And yet my account grew by a number I could count on one hand.

Something was broken. I just couldn't see it.

I'd been tracking daily analytics in a CSV. Every morning, pulling numbers from X's dashboard like a mechanic reading engine diagnostics. Impressions. Likes. Follows. Unfollows. I had the data. What I didn't have was the right question.

Then I calculated something I'd never tracked before.

Profile visits ÷ impressions.

On November 10, that number was: 26 ÷ 9,930 = 0.26%.

Twenty-six people. Out of nearly ten thousand who saw my content, twenty-six thought to check who made it. And here's what made it worse: of those 26 who visited my profile, 6 followed. That's a 23% conversion rate, which is actually solid. My profile was doing its job. My bio, my pinned post, my visual grid. All working.

The pipeline wasn't leaking at the bottom. It was bone dry at the top.

I started calling this number the curiosity rate: the percentage of people who see your content and get curious enough to click through to your profile. Platform benchmarks sit between 1% and 5%. I was at 0.26%. Bottom of the barrel. Not even close.

Here's what that meant in real numbers:

I was leaving 17 to 63 followers on the table. Every single day.

Not because my content was bad. Not because my profile wasn't converting. Because nothing in my posts gave anyone a reason to wonder who was making this stuff.

If you're building AI art, whether it's Firefly, Midjourney, Stable Diffusion, whatever. You've probably noticed something. Your images get engagement. People double-tap. They save them. They might even share them. But the follow doesn't come.

Here's why, and it took me embarrassingly long to figure out.

When someone scrolls past a beautiful AI-generated image, they engage with the image. They bookmark the image. The image is the complete experience. There's no lingering question of "who made this and what else do they have?" Because the image already gave them everything.

Compare that to someone who posts a thread breaking down their creative process, the prompt iterations, the failures, the technique that finally cracked the look they wanted. That content creates an open loop. The reader finishes and thinks, "What else does this person know?"

My November 10 content was gorgeous. Beautiful surrealist scenes. People appreciated them. And then they scrolled on, because the images answered every question they had.

The content was complete. Too complete. There was nothing left to be curious about.

November 11. One day later. I changed one thing.

I'd been closing every post the way you're told to: "Like, share, bookmark and follow for more." The generic call-to-action that every growth guide tells you to use. It's wallpaper. Everyone says it, nobody reads it.

I replaced it with something specific:

"Technique #3 of 47 I've documented. Full library on profile."

That's it. Instead of asking for engagement, I hinted at a library. Instead of begging for a follow, I implied there was more to find. The post became a window into a larger body of work, not a standalone piece.

The math changed overnight.

Look at Nov 11. Fewer impressions than the day before. Same number of profile visits, 26 both days. But the follows jumped from 6 to 32. Five times more followers with less reach.

The curiosity rate ticked up slightly, but something more interesting happened. The people who did visit were far more likely to follow. The CTA language wasn't just driving traffic. It was pre-qualifying visitors. By promising a library of documented techniques, I was filtering for the people who actually cared about process. Those people land on a profile full of educational AI content and think, "Yes, this is exactly what I need."

Over the next five days, my net growth hit +97. The previous five days: +39. A 149% increase from changing a single line of text.

That CTA fix opened a door into a much bigger realization. Once I started tracking which content types actually drove follows, not just engagement, not just impressions, but actual "I'm going to follow this person" decisions. The results were counterintuitive.

I analyzed all 1,789 posts from the year. Here's what the conversion data looked like:

Shoutouts (recommending other creators): 1.08 follows per post Stor-AI content (my AI storybook series): 0.35 follows per post Algorithm threads (educational breakdowns): 0.33 follows per post Tool showcases (pure image posts): 0.05 follows per post Tutorial/Process posts: 0.02 follows per post

Read that again. Community shoutouts, posts where I'm literally telling people to follow someone else... converted at 3x the rate of my next best category. Meanwhile, my pure tutorial content, the stuff I spent the most time crafting, barely moved the needle.

Why? Because shoutouts signal community. They tell people "this account connects you to other interesting creators." That's a follow-worthy promise. A single beautiful image, no matter how technically impressive, isn't.

And the statistical correlations backed this up across the entire active period:

You don't grow by posting more. The number of posts I published on any given day had almost zero statistical relationship with how many followers I gained. What mattered was whether the content made people save it and check my profile. Quality signals, not quantity.

By mid-November, I'd identified a three-part combination that consistently produced 30+ follower days:

One: An educational thread in the morning. Not just "look at this image" but a systematic breakdown of method. Prompt structure, tool comparison, iteration process. The kind of post where someone thinks "I need to see what else this person has documented."

Two: A strategic amplification reply to a larger creator. Not a "great post!" comment, but a genuine addition to their conversation, with a visual example. This put my work in front of new audiences who were already primed to engage with AI content.

Three: Community foundation: 50+ daily replies to real conversations. Not drive-by engagement. Actual conversations.

When all three were present: average +31 to +41 net growth per day. When only one or two hit: average +12. When none: +4. Like November 10.

The transformation from dormant to active was a 16.7x increase in daily follower growth. From +0.9/day during the first eight months to +15.0/day during the active growth period. And it wasn't because the art got better. The images I was creating in March were just as surreal, just as technically accomplished as the ones in December.

What changed was the framing.

If you're an AI creator watching your engagement stay healthy while your follower count flatlines, here's the diagnostic:

Step one: Calculate your curiosity rate. Pull today's profile visits, divide by impressions. If you're below 0.5%, your content isn't creating open loops. People are enjoying it and moving on.

Step two: Reframe your CTAs. Stop asking for engagement. Start hinting at depth. "One of 12 techniques I tested this month" beats "follow for more" every time. Your audience should finish a post with an unanswered question: what are the other 11?

Step three: Track your follow-per-visit rate. If people visit your profile and don't follow, that's a profile problem. If they never visit, that's a content framing problem. Different diagnosis, different fix.

Step four: Check your content mix. If everything you post is a finished image with no context, you're giving people the dessert without the menu. Mix in process, mix in community, mix in documentation of your method. The luthier in me (I spent 12 years building guitars) learned this a long time ago: people don't just want the instrument, they want to watch it being made.

My bookmarks grew 85x over the year. My followers grew 31.5x. Both numbers moved because I stopped thinking of each post as a standalone performance and started treating every piece of content as one page in a visible, findable library.

The work didn't change. The signal did.

Glenn is an @AdobeFirefly Ambassador and AI creator documenting the craft of prompt engineering and creative process at @GlennHasABeard. He publishes The Render newsletter and is currently 1,390 followers away from his 5,000 goal.

This article is part of 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.