When it feels like your illustrations aren't getting traction, the problem usually breaks down into one of three things: the post isn't being shown at all, it's being shown but nobody stops scrolling, or people stop but have no reason to react. These three have completely different causes and completely different fixes, so as long as you lump them together as "my art isn't growing," your efforts will keep missing the mark. In this article, we use facts you can verify in the recommendation code X has made public to show how to pinpoint the cause, along with concrete fixes you can start trying today.

A word of caution up front: nothing here is a guarantee that "do this and your posts will take off." Parts of the distribution system are not public, and results also depend on the artwork itself and on timing. What this article can offer is limited to "ways to avoid being at a disadvantage" and "facts that can be verified."

Note: Figmee currently offers figurine-style image generation, 3D model data downloads (GLB / 3MF), and turntable video export. Ordering physical 3D prints is Coming Soon.

Break "not growing" down into three stages

When you look at your numbers, separate them into three stages: impressions (how many times your post was shown), dwell (how long people looked at it), and engagement (likes, replies, reposts). Which stage is the bottleneck determines what you need to fix.

  • Stuck at stage 1: Impressions barely move. At this stage, the issue is likely not the quality of your work but that your post isn't making it into distribution in the first place.
  • Stuck at stage 2: You're getting impressions, but engagement is extremely low. Either the first image isn't making people stop scrolling, or it stops them but doesn't get its message across.
  • Stuck at stage 3: People are seeing your work, but they have no motivation to react. There's no room for them to leave a comment, or nothing for them to look at next.

Even under the same "not growing" label, a stage 1 problem calls for designing your reach rather than your presentation, and a stage 3 problem is about the path you give viewers rather than the artwork. Mix these up, and you'll keep fixing things that didn't need fixing.

What the recommendation system lets us verify

X has published the code for the recommendation algorithm that powers the For You feed (xai-org/x-algorithm, Apache-2.0 license). The facts you can read from it differ quite a bit from the conventional wisdom that circulates online.

There is one important caveat, though. The numeric weights for each signal are not included in the code. They are configuration values fetched externally at runtime, so while you can see the sign of each signal (whether it adds to or subtracts from the score), you cannot see how much each one matters. Read everything below as a discussion of signs, not sizes.

What people often sayWhat the public code lets us verify
Posting external links lowers your reachNo logic that penalizes links can be found. In fact, copying a link to share it exists as a positive signal
If you have few followers, you won't be shownFor distribution beyond your followers, candidates are chosen by how close interest embeddings are, and there are paths that don't use follower count as a direct criterion
Images do better than video (or the other way around)Neither is penalized. For videos above a certain length, viewing quality counts as a positive; for images, being expanded counts as a positive
Older posts gradually lose reachNo logic that lowers scores continuously over time can be found. Instead, posts past a certain age are simply cut from the candidate pool
Posting in rapid succession increases your reachWhen the same author appears repeatedly within a single feed load, later posts are dampened. Only the single best post from the same conversation remains

The code also spells out what clearly sits on the negative side. Posts predicted to trigger "Not interested," "Mute," "Block," or "Report" have their scores pushed down. In addition, posts predicted to be shown but not dwelled on are also on the negative side. Conversely, dwell time is treated as a positive that counts twice.

This isn't unique to X, either. Instagram has officially explained that it uses different signals for Feed, Reels, and Explore, and that what a person has interacted with in the past carries a lot of weight (source: Instagram, "Instagram Ranking Explained", as of September 2026). Across platforms, "closeness of interests" and "dwell and engagement" are the shared core.

Fixes for stage 1: getting into the game

These are the moves that work when your posts simply aren't being shown.

  1. Set your account to public (about 1 minute) — If your account is private, its content is excluded from understanding and annotation, which makes it hard to be discovered by people who don't follow you. This is a prerequisite.
  2. Keep your themes and vocabulary consistent (about 30 minutes) — Distribution beyond your followers is driven by closeness of interests. If your art, casual chatter, and promotion are all over the place, it becomes unclear which interests a post is meant for. Narrow your account down to one or two main themes.
  3. Stop posting in rapid bursts (0 minutes) — Consecutive exposure from the same author is dampened, and only the best post from a conversation survives. Five posts in a row won't get you five times the reach. Make each post self-contained, or gather them into a single thread.
  4. Post at times when people react (about 10 minutes) — The posting time itself earns no bonus, but posts drop out of the candidate pool after a certain age. That means posting when you can get early momentum does matter. Check which times your past posts got reactions quickly. We walk through how to find your time slots, step by step with a tracking sheet, in What Is the Best Time to Post Illustrations?.

Fixes for stage 2: stop the scroll with the first image

This is the stage where your post is being shown but getting few reactions. Here, the issue is often not the quality of your art but the first image and the first two lines.

  1. Put the payoff in the first image (about 15 minutes) — Make it clear what the piece is and what's worth looking at, without anyone needing to tap to expand. If the first image is a work-in-progress photo or a rough sketch, most people won't stop there.
  2. Include details that make people want to expand the image (about 15 minutes) — An image being expanded is a positive signal. Fine detail, before-and-after comparisons, and two images side by side all give people a reason to tap in.
  3. Skip the preamble in the first two lines (about 5 minutes) — Starting with "Long time no see" or "Sorry, it's not very good" gets your post scrolled past before people reach the point. Start by talking about the artwork.
  4. Use something with motion as the first item (a few minutes or more) — For videos above a certain length, viewing quality counts as a positive. But a video that is long and loses viewers halfway through backfires, so the baseline is to put the highlight at the start and keep it short enough to be watched to the end.

Fixes for stage 3: give people a reason to react

This is the stage where people are seeing your work but not reacting. Replies, quotes, profile clicks, and follows all exist as positive signals. In other words, whether you provide an "entry point" for reactions is something you can control.

  1. Add one line with a question or some open space (about 3 minutes) — Short, easy-to-answer questions work well, like "Which color scheme do you prefer?" or "What should I draw next?" Asking for long feedback raises the bar for replying.
  2. Share one thing about your process (about 10 minutes) — Where you hesitated, ideas you dropped, how long it took. That gives people more of a hook to start a conversation than the finished piece alone.
  3. Tidy up your profile and pinned post (about 30 minutes) — The next thing someone who likes your post looks at is your profile. Make it clear within three seconds what kind of work you make.
  4. Don't send promo-only replies to other people's posts (0 minutes) — Mechanical replies from accounts with few followers fall into territory that can be flagged as spam. You risk tripping negative signals in exchange for short-term exposure.

Expanding how you present your work

If you've tried all of the above and nothing moves, the remaining direction is "showing the same work in a different form." Rather than remaking the artwork, the idea is to add more channels for getting it across.

One of those is turning it into 3D. If you create a 3D model from your illustration, you can post it as a rotating video a few seconds long. A flat illustration starting to spin is itself something that catches the eye mid-scroll. Another advantage is that information that sat outside the frame in a still image, like thickness, the sides, and the back, comes across all at once.

That said, this is not magic. Turning it into a turntable video doesn't guarantee more reactions. If it works, it works purely as an additional tool: people stop for a moment because it's unusual, you get to post the same piece twice, and you have a 3D angle to start conversations with. There's also a side effect: if the back of the model isn't well made, that becomes visible the moment it rotates.

The process goes like this: upload your art, generate a figurine-style image, generate a 3D model from that, and check it in the browser while rotating it 360°. With Figmee, you can export a turntable video from the generated 3D model on the spot, and you can also download it in GLB / 3MF format. Image generation is free for the first 5 times as a sign-up perk (valid for 3 months from registration, first sign-up only), and 3D model generation is free for your first model per account, then 5 tokens per model after that ($5.99 for 5 tokens). For the full picture, see How to Turn an Illustration into 3D (2026 Guide); for how to make videos for posting, see How to Turn a 3D Model into a Turntable Video for Social Media.

Beyond 3D, there are other options, such as animating your art or changing where you show it. For a comparison of methods, see How to Animate an Illustration or Turn It into 3D.

Have a place ready to catch the reactions

Another thing that's easy to overlook is where people go next after seeing your post. Profile clicks and follows exist as positive signals, but if your work isn't gathered in one place at the destination, the momentum stops right there.

Posts flow past, so it's not realistic to expect people to scroll back through your older work. That's exactly why having a separate place where your work can be seen together on one page changes what you get out of a single post. With Figmee, if you turn on the toggle in My Page to make it public, a creator page with a public profile and a grid of your works becomes visible, and others can follow you (it's private by default). You can also display your works in the public gallery.

Your own website or an external portfolio service — any location is fine. What matters is that there is "one page for people who liked your work to look at next."

A caution when looking at the numbers

Finally, one point about mindset. The numbers for each individual post are separate from the value of your work. Reading the public code makes it clear that whether a post is shown depends heavily on how close it is to what the viewer has reacted to in the past. In other words, the same piece can get different results depending on who it reaches.

With that in mind, judge improvements in units of 10 posts, not one. If you change your style or posting frequency in response to a single result, you'll never know what actually worked. Changing one thing at a time is, in the end, the fastest route.

Frequently asked questions

Q. Does putting an external link in the post lower my reach? A. Within the scope of the public code, no logic that penalizes external links can be found. In fact, copying and sharing a link exists as a positive signal. However, a link card can weaken how the first image comes across, so the safer structure is to draw people in with media first and then place the link.

Q. Is everything pointless while I still have few followers? A. Not necessarily. Candidate selection for distribution beyond your followers has a path that uses closeness of interests rather than follower count, so in principle you can be picked up even with few followers. On the other hand, there are also auxiliary paths where follower count may play a role, and how much they matter can't be confirmed from public information. What's certain is only this: keeping your theme consistent so your posts reach people with similar interests won't put you at a disadvantage.

Q. How many times a day can I post? A. There's no fixed limit on the number of posts, but there is a mechanism that dampens later posts when posts from the same author line up within a short time. Also, only the best post from the same conversation remains. The system is structured so that focusing on the quality of each post fits it better than pushing volume.

Q. Should I add hashtags? A. No logic that penalizes adding them can be found. However, lots of tags unrelated to the content risk pushing toward "Not interested" or mutes. There is also logic that excludes posts matching muted words from the candidate pool, so it's safer to avoid lining up words unrelated to the topic.

Q. Will making a 3D model increase reactions? A. Not necessarily. Turning your art into 3D "adds one more way to get people to stop"; it doesn't guarantee reactions. If it has an effect, it's that thickness and the back, which don't come across in a still image, can be conveyed, and that you can put out the same piece again from a different angle. Use it as one more option for presentation, on the premise that the appeal of the artwork itself is the foundation.

Summary

When you feel your illustrations aren't getting traction, start by separating where you're stuck: "not reaching people," "not stopping them," or "not prompting a reaction." From the public recommendation algorithm code, you can verify that there's no penalty for external links, that there are distribution paths that don't directly use follower count, that rapid-fire posting makes your own posts compete with each other, and that not being dwelled on sits on the negative side. On the other hand, the weights for each signal are not public, so that's as far as anything can be stated with certainty. Trying one change at a time and checking in units of 10 posts may look like a detour, but it's the fastest way.