The slideshow play.

What it is, in one paragraph.

You describe a small consumer app to Claude Code in three lines and let it build the whole thing in one sitting: the screens, the paywall, the store listing. You submit it. While Apple reviews, you open a fresh TikTok account and post photo-mode slideshows, two or three a day, three to six slides each, in the voice of an ordinary person who happens to use the app. Most posts do nothing. One in fifty catches, does a few hundred thousand views, and sends a thin stream of people to the App Store. If the app charges money and the stream is thin but real, you have a business. If not, you ship the next app next week and post again. The unit of work is the week, and the unit of luck is the post.

It is a volume play in both halves. That is its strength, and, as of this summer, exactly what both TikTok and Apple have tuned themselves against. Hold both thoughts.

Why it works, according to the people it is working for.

1. Slideshows out-reach video right nowPhoto mode makes the viewer swipe, and every swipe is an engagement signal. Tools in the space claim five times the reach of a video from the same account, and some creators report ten. Treat the multiple as marketing; treat the direction as true. Every account we looked at that runs both formats says the same thing: slides travel further than clips.
2. Certain formats keep catching, across every nicheAmos Bastian studied the slideshows behind the “thousands in MRR” posts on X and found ten shapes that repeat: tips and mistakes, emotional reveals, ranked roundups, problem-and-checklist, interactive “pick one” posts, before-and-after, relatable one-liners, audio showcases, personal timelines, and news. The interactive shape was the outlier: his top five added up to 221.5 million views. His main finding is the one to keep: the app does not need to be the entire post. It appears inside a story someone would have watched anyway.
3. The whole pipeline can be automated for pocket moneyA developer called ashen took a vibe-coded Quran app to its first hundred downloads and a thousand likes without touching a post: an AI agent writes and renders three slides, Postiz schedules them as drafts at $29 a month, and a $50 Android phone does the final tap so TikTok sees a human device. Maintenance is about an hour a month. A separate Claude Code skill, TikTok App Marketing, runs research, image generation, posting, and a daily analytics cron on the same pattern, and its author claims over a million TikTok views and $670 a month of MRR from an agent on an old gaming PC.
4. It scales to real numbersThere is a full walkthrough on YouTube titled “100,000 App Downloads From TikTok Slideshows,” and the Couple Joy account, which the Playkit Report broke down, sits at two million views and sixty thousand likes over two hundred posts, all faceless. The X guides that get passed around, Sulfur’s complete guide and Lino Leighton’s “easy mode,” both start from the same premise: this is the easiest organic channel a consumer app has ever had.

The four things that quietly kill it.

1. 1. Viral is not revenueMax on X, in a thread that gets reposted every few weeks: when you see a viral slideshow for an app, look up the app’s downloads and revenue before you copy the post, because the two often do not match. The arithmetic behind that: a good post sends roughly 0.2 to 0.5 percent of viewers to the bio link. Five hundred thousand views is one to two and a half thousand taps. Then the App Store page has to convert, then the paywall has to convert, and each step halves you. Slideshows also earn exactly nothing from TikTok’s own Creator Rewards; photo mode is excluded. The app is the only way the views become money.
2. 2. Most apps never get there anywayRevenueCat’s State of Subscription Apps, from tens of thousands of real apps: roughly one in five reaches $1,000 in revenue within two years and about one in twenty reaches $10,000, and the drop-off roughly halves at every milestone after that. Frank Andrade, who writes about making money with Claude, puts the whole category bluntly: high margins on paper, getting users is brutal, and someone copies you next week. Slideshows fix the first sentence and none of the others.
3. 3. TikTok now punishes the way the play is usually runThe 2026 triggers for reduced visibility are, in order, reposted watermarked content, undisclosed AI content, banned hashtags and automation behaviour. Several accounts on one device or one IP get throttled together, and a flagged device can carry the flag to a new account. Every serious guide, including the automation ones, now says the same three things: warm a new account for one to two weeks of ordinary human use, post from a real phone, one account per device. The people who lost accounts skipped one of those.
4. 4. Apple closed the door on the one-shot template in JuneOn June 8, 2026 Apple rewrote guideline 4.3(b) to reject “mediocre, low-quality, or low-effort” apps that “do not add value,” naming the categories it will now refuse outright unless the app is meaningfully different: dating, flashlight, sound effects, wallpaper, timers, fortune telling. The trigger was an 84 percent surge in submissions it attributed to AI coding tools. In practice 4.3 is a judgment call about originality, it is the hardest rejection to appeal, and repeated submissions of the same kind can now cost you the developer account. A chat box wired to an LLM and a habit tracker from the same prompt everyone used are the two shapes reviewers say they spot instantly.

Put the bulls and the bears together and the picture is not “it doesn’t work.” It is: the reach is real, the conversion is thin, and the lazy version of both halves is now actively penalised. So do the non-lazy version.

How to run it in September 2026.

1. Build for one screenshotBefore you prompt anything, write the single frame that would stop a thumb: a result, a before-and-after, a verdict, a list the viewer wants to argue with. If the app has no such frame, it is not a slideshow app, pick another. Then hand Claude Code the three lines and let it one-shot the build, but give it one native thing a template would not have: a widget, a camera flow, offline data, a real integration. That one thing is also your 4.3(b) answer.
2. Charge from day onePaywall in the build, RevenueCat wired, weekly and annual. You are about to spend a month generating a thin stream; the stream has to hit a price or you learn nothing. Subscriptions through Apple, not Stripe, or that is a 3.1.1 rejection before anyone sees a slide.
3. Warm the account before you need itOpen the TikTok account the day you submit to Apple, on a phone that is only for this account. Use it like a person for one to two weeks: watch, like, comment, save, follow twenty accounts in your niche. Do not post. Review takes that long anyway.
4. Pick three series, not three postsFrom the ten shapes above choose three that fit your frame and commit to them as recurring series: an interactive “which one are you”, a tips-and-mistakes, a before-and-after or personal timeline. The first slide names one specific situation without exaggerating it. The app shows up on slide three or four as the thing the narrator used, never as the subject. Caption and sound are chosen by hand; sound choice is the part every automation guide admits it cannot do.
5. Two or three a day, from the phoneLet an agent draft the slides and captions if you like; the last tap is human, on the device, with a trending sound. One account, one device, one app. If you want a second account it gets its own phone and its own network. This is the entire difference between the people who kept their accounts and the people who did not.
6. Measure the thing that paysViews are the vanity layer. Track four layers separately: saves and comments, profile taps and link taps, installs in your own analytics, and paid conversions in RevenueCat. Judge a series on layer three and four after thirty days. A post that did a million views and eleven installs is a failed post with good lighting.
7. Thirty days, then a decisionIf a series is producing installs, double its cadence and cut the other two. If nothing is producing installs after thirty days and roughly seventy posts, the frame was wrong, not the effort. Ship the next app. That is not a failure of the play; it is the play. The people at real numbers are on their fourth or fifth app, and the previous ones are still in the store, still earning a little.

What one landed post is actually worth.

Take the bears’ own numbers and be generous with them. A post that catches and does five hundred thousand views sends, at the top of the quoted range, about two and a half thousand people to your link. Say a third of them install, which is good for a cold App Store page. That is eight hundred installs. Say one in twenty starts a trial and half of those pay: forty paying people, at a typical five to eight dollars a week or forty to sixty a year. One landed post is a few hundred dollars a month of MRR, decaying as they churn, and you will not know which post will land in advance.

That is the whole reason the play is a volume play. Ten landed posts across a year, on two or three apps, is the “thousands in MRR” that the X threads are describing, and nobody in those threads got there with one app or one week. If you want to know whether the play is worth your time, that is the number to hold against your alternatives, not the seven-million-view screenshot.

Plainly. We used the word “spam” in the title of the note this page grew from, and we are keeping it here because it is the honest word for the volume half. But the version of spam that TikTok and Apple now catch is the lazy one: same template, same prompt, same device, undisclosed AI images, posts that are ads. Run the disciplined version above and it stops being spam and starts being a publishing schedule. Same effort, different outcome.

Where every number comes from

• Amos Bastian, Indie Hackers, July 2026: I had no idea TikTok slideshows could get apps to 1000s in MRR. The ten formats and the 221.5M-view figure. The author sells a slideshow tool, Autovirality.
• Postiz blog: How a developer fully automated TikTok slideshow marketing with AI agents. Three posts a day, $29 Postiz, $50 Android phone, 100+ downloads. Postiz sells the scheduler.
• Claude Code skill: tiktok-app-marketing. The 1M+ views and $670/month MRR claim; the 7-to-14-day warm-up and manual-sound caveats are the author’s own.
• Max on X: viral slideshows ≠ revenue. Also Sulfur’s complete guide to TikTok slideshows and Lino Leighton’s highly converting slideshows, easy mode, both behind X’s login.
• OpenClip: Do TikTok slideshows make money? The 0.2–0.5% bio-link rate and the Creator Rewards exclusion.
• RevenueCat, State of Subscription Apps 2025: ~20% of apps reach $1,000 and ~5% reach $10,000 within two years.
• Frank Andrade, Artificial Corner: How to make money with Claude in 2026. The “getting users is brutal” line.
• 9to5Mac, June 9, 2026: Apple tightens App Store rules against apps that “do not add value”. The 4.3(b) wording, the named categories, and the 84% surge. Also Why your AI-built app won’t pass review in 2026 on DEV for the seven rejection patterns.
• Shadowban triggers and the multi-account rule: Shopify, Multilogin and InfluenceFlow, all 2026.
• Reach multiples: ReelBase on the photo-mode algorithm, which sells a slideshow tool and cites no outside data; it also warns the window is six to twelve months before saturation. AttentionClaw for the four measurement layers.
• Scale examples: 100,000 App Downloads From TikTok Slideshows on YouTube, and the Playkit Report on Couple Joy’s faceless slideshow growth, paywalled past the headline figures.

Nothing above has been audited by Forma. Several sources sell the tool they are describing and are marked as such. The arithmetic section is an illustration built from their quoted rates, not a forecast. Claude Code is Anthropic’s; TikTok is ByteDance’s; the App Store is Apple’s; none of them endorse this page.