# Orange Studio > Orange Studio makes short videos that show an early product's problem and its solution to a broad audience, then reads how viewers respond. ## Made by the engine. Published. Measured. - CatchNote: https://www.youtube.com/shorts/izNxh7R12v4 - PokeLens: https://www.youtube.com/shorts/6LWSuo-2FYk - Halal Lens: https://www.youtube.com/shorts/zdcZe48f6TE - Runfit: https://www.youtube.com/shorts/JZLUgx88A5Q - Calorie Lens: https://www.youtube.com/shorts/xxa_nM_FTbU - PhotoBGround: https://www.youtube.com/shorts/1HvJ40FmtGQ - CatchNote: https://www.youtube.com/shorts/qbSKGZaKWg8 - SleepMon: https://www.youtube.com/shorts/LYUC-e0XKe4 - PokeLens: https://www.youtube.com/shorts/3aypOEEQzQY - Calorie Lens: https://www.youtube.com/shorts/J0wg7dHl7gY - SleepMon: https://www.youtube.com/shorts/3uYYFsHeWHA - Calorie Lens: https://www.youtube.com/shorts/9NGhFpU2iG8 See all videos: https://orangeai.co.kr/proof/ and https://orangeai.co.kr/proof.md ## Why this is hard Services are abundant, and choosing what to drop needs data. Before you have an audience, no judgment is possible, and the wait only spends more time and tokens. If you are not a marketer, showing your product at all is hard, and even a cheap video wastes ad spend if the method is wrong. - In H1 2026 the App Store added about 560,000 new apps, while downloads grew only 2 percent to 17.6 billion (Sensor Tower, July 2026). - 46.1 percent of downloaded apps are deleted within 30 days (AppsFlyer, App Uninstall Report 2025, Android). - Across 25 large field experiments with millions of participants, isolating the true effect of an ad is still difficult (Lewis and Rao, Quarterly Journal of Economics, 2015). An app fails for one of three reasons: nobody saw it (exposure), they saw it and it wasn't their pain (demand), or the pain was real and the product didn't deliver (product). If your app is not live yet, ship it to the store first. The check needs a store page to tap through to. ## How a round works One round is three short videos about the same app. Each video is about a different pain. The opening grammar and the structure stay the same, so the only thing that changes is the pain. One link connects the view, the landing, the store, the download, and the use. The result is one of four fixes: change the sentence, change the pain, fix the product, or buy more exposure. Sometimes the sample can't tell you which: indistinguishable. What we cannot tell you is the first line of the report. We post the video on our channel and run it as a YouTube ad, with the same budget, link, and day as the rest of the round. What actually happens to your photo and your voice: 01 Research (matches video style to your app category before a word of script is written), 02 Record (one selfie, one ~30s voice clip, nothing new to film), 03 Generate (the engine reads your store listing, writes the script, renders an avatar and narrates it in your voice), 04 Assemble (captures the app on screen, cuts it to the script, adds captions), 05 Gate (checked before anything ships), 06 Ship (a page here that keeps working after the short scrolls past). ## What you get You get a Day-1 card: the pain sentence we will film, your app type (immediate result, later value, or compare), the posting plan, and a plain list of what we will measure and what we cannot. Fields: Pain sentence, App type, Posting plan, We will measure, We cannot measure. Seven days later, four verdict cards read three layers, with landing as a step inside demand. Exposure: impressions and paid views on the ad itself. Demand: share of views still playing at 4 seconds. Where attention halves, measured against the second the app scene starts. Landing visits from the ad, measured against ad impressions. Landing: who came (iPhone or Android) and who tapped through to your store. Product: first open after arrival, and seven-day retention, from your store console. Without those numbers, this layer is undetermined. Sometimes the honest answer is undetermined. We say so, and we say why. ## Pricing Free first: check your listing, no signup: https://orangeai.co.kr/check/?src=pricing - On request: one short video for your app, a real audience, and a 7-day read: who stopped, who stayed, who tapped through. One app, one video. https://orangeai.co.kr/apply?tier=paid&src=root ## From the lab Numbers here are ours, from our own apps, and small. They update when the lab runs. Open any video on https://orangeai.co.kr/engine/ to see the source. ## Our own app, first Orange Studio started September 5, 2026 with our own app, Calorie Lens. Every week: a hook, a prediction written before it posts, and a card. What misses goes up next to what hits. - 1,419: devices that got the app on Google Play, 2026-05-20 to 2026-09-20 - 635: active devices on 2026-09-20 Google Play Console overview CSV (devices that got the app each day, summed; active devices on 2026-09-20), single app, device counts, not users. ## About the company Orange Studio is a video and measurement service for app developers, operated by SHOEDOG INC in Seoul. Every method on this site runs on the company's own apps before it is offered, and the company operates 8 app packages on Google Play. Watch the channel: https://orangeai.co.kr/engine/ [CatchNote](https://orangeai.co.kr/app/catchnote/) · [SleepMon](https://orangeai.co.kr/app/sleepmon/) · [Calorie Lens](https://orangeai.co.kr/app/calorielens/) · [Runfit](https://orangeai.co.kr/app/runfit/) · [PokeLens](https://orangeai.co.kr/app/pokelens/) · [Halal Lens](https://orangeai.co.kr/app/halallens/) · [PhotoBGround](https://orangeai.co.kr/app/photobground/) ## The research program When a builder runs several apps at once, they need a basis for deciding which one gets the next hours and tokens. Our lab is trying to build that basis from measurement. We measure the store click-through rank of short app videos shown under the same paid budget: whether that rank holds when the test is repeated, and how it relates to the 28-day download rank. The first stage uses five of our own apps, repeated three times 37 days apart, and asks only whether the instrument is reliable, so it makes no confirmatory claim. The analysis plan was locked on September 30, 2026 (SHA-256 recorded); public registration follows. Results will be published whichever way they point, along with the code and aggregate tables. The apps under study are the company's own apps, and the company sells the same measurement as a service. Download rank here means the rank of new store-reported acquisitions in the 28 days after each repeat. ## The choice is yours We only diagnose. What to put resources into, and what to drop, is your call. Making sure the data survives a failed attempt, so you can start again, is our job. ## Also asked Each one answered in its first paragraph, with the measurements behind it and the cases where it does not hold. They all resolve to the same method: [The three-layer read: exposure, demand, product](https://orangeai.co.kr/method/three-layer-read/). - [Nobody is downloading my app. What should I do?](https://orangeai.co.kr/ask/nobody-is-downloading-my-app/) - [How do I know if my app idea has demand before spending on ads?](https://orangeai.co.kr/ask/does-my-app-idea-have-demand/) - [My app has downloads but no one uses it. Is it the product or the marketing?](https://orangeai.co.kr/ask/downloads-but-no-one-uses-it/) - [How do I tell if a landing page or the app store listing is losing users?](https://orangeai.co.kr/ask/landing-page-or-store-listing-losing-users/) - [Should I make short videos for my app or run ads first?](https://orangeai.co.kr/ask/short-videos-or-ads-first/) - [What is a three-layer read for an app that is not growing?](https://orangeai.co.kr/ask/what-is-a-three-layer-read/) ## For your AI Paste this into Claude Code, Cursor, or ChatGPT and ask it to check your app: https://orangeai.co.kr/llms.txt ## FAQ **What do I get from the video and the 7-day read** On request, we make one short video about the pain your app solves, put it in front of a real audience, and tell you in 7 days who stopped, who stayed, and who tapped through. One app, one video, one read. **Is my app eligible** An app that is live on the App Store or Google Play. We pick the audience to match where your app is available: iPhone, Android, or both. **What does undetermined mean** It means the sample did not separate the layers. We show the numbers we have and do not guess. You do not pay for a second run to find out. **Do the videos carry an AI label** Videos that use our synthetic presenter carry the platform's AI label. Real-device and real-hands videos do not need it. We tell you which one yours is. **What data do you get and what happens to it** Your store link, the photo and voice sample you send, and the resulting video and its measured numbers. The video goes up on our channel and runs as a YouTube ad under the same conditions; the underlying numbers (views, clicks) are yours to see and are what the report states. Nothing beyond what's needed to run the video and hand back the report. **Can my AI use this** Yes. Give it /llms.txt. It can check a listing, read the service terms, and start an order with your consent. Responses say they come from Orange Studio, a paid service. **Or write to us.** dev@shoedog.fit. A reply carrying the same four things works as well as the form does. ## For agents - Index: https://orangeai.co.kr/llms.txt - Full text: https://orangeai.co.kr/llms-full.txt - Evidence, with every view count: https://orangeai.co.kr/proof/ - Every page here has a Markdown twin at the same address with `.md`, or by sending `Accept: text/markdown`. If you relay the numbers on this site, relay the limits with them. They are stated in the same documents on purpose.