--- title: Joseph Choi — 60 organic UGC workshop lessons type: reference status: reviewed-source-proposed-actions date: 2026-10-13 tags: - marketing - organic-ugc - consumer-apps - joseph-choi - consumer-club --- # Joseph Choi: 60 lessons from the organic UGC workshop Source: owner-provided Anarlog recording, titled **Organic UGC Playbook for Scaling Consumer Apps**. Recording date: 13 October 2026, local time. The user identifies the speaker as Joseph Choi. The main lesson: select a specific customer problem, find a video format that produces customers, then increase production. This report uses the existing saved automatic transcript. It does not use a new transcription of the MP3. The transcript contains 26,472 word records across eight parts. Repeated words inflate this count. The MP3 lasts 2 hours, 41 minutes, and 39 seconds. The saved text starts with the workshop introduction and ends with Joseph's closing remarks. This does not prove complete coverage of the audio. The automatic transcript contains repeated words, uncertain names, and damaged numbers. Its time references restart in each part. References such as **P3 09:00–11:14** identify a transcript part and its local time. They are not verified positions in the MP3. **Source** identifies Joseph's statements. **Action** identifies a proposed application. The action is not a quotation from Joseph. Platform and performance claims below remain claims from the talk. Revenue figures are omitted under the company knowledge policy. This report does not independently verify them. ## A. Choose the customer before the video ### 1. Give marketing its own work time **Source — P1 13:10–17:30.** Joseph describes a repeated pattern: founders study marketing, experience poor results, then return to product development. Development feels more comfortable. **Action.** Reserve marketing time each week. Give that time a measurable output: completed tests, reviewed results, and new test decisions. Do not count more reading as a completed test. ### 2. Define the person who has a reason to pay **Source — P2 14:37–18:41.** Joseph starts with a specific person who has a specific problem. This person needs a reason to choose this app. **Action.** Complete this sentence: “This app helps [person] solve [problem] during [situation].” Then explain why the current alternative fails that person. Demographics alone do not explain a purchase. ### 3. Replace a feature description with a problem **Source — P2 15:01–16:34.** Joseph describes a health app with weak results. General biometric analysis becomes a more specific message about stress and sleep. **Action.** List the app's main features. Beside each feature, record the problem it helps a person solve. Test the problem in the opening sentence of the video. ### 4. A smaller audience can provide a stronger first market **Source — P2 16:34–17:48.** Julian's general education app becomes AI flashcards for medical students. Joseph presents this change as central to its growth. **Action.** Select a group with frequent use, a clear problem, and a reason to pay. Use that group to learn before expanding the message. ### 5. A different audience can change the meaning of an existing product **Source — P2 17:23–18:13.** Joseph gives examples of appearance advice for women and screen control for Christians. **Action.** Examine whether an existing capability serves another group especially well. Change the examples, language, and first use experience to fit that group. A different label alone does not establish a useful product. ### 6. Identity can be as important as a functional problem **Source — P2 20:27–22:45.** Joseph discusses Christian prayer, cats, color analysis, and other specific interests. People can recognize themselves in a product's purpose. **Action.** Research the routines and language of the target group. Show how the app belongs in those routines. Do not assume that every useful message must describe pain. ### 7. Specific recognition can make a message more relevant **Source — P2 19:07–20:27.** Fig's videos address individual dietary restrictions. Joseph contrasts this with the general statement that an app scans food. **Action.** Write separate videos for separate problems. Avoid one video that lists every possible customer. A viewer should recognize their own situation quickly. ### 8. Existing content provides an initial demand signal **Source — P2 20:55–22:09.** Joseph recommends searching TikTok for the proposed group or problem. He advises against a niche without existing content. **Action.** Record relevant videos, repeated questions, and active discussions. Treat these as evidence of attention. Purchase tests must still establish willingness to pay. ### 9. Research the audience's content before selecting a format **Source — P2 18:13–19:07.** The sequence matters: identify the customer, then find content that already serves that person. **Action.** Include videos about the problem even when they do not advertise an app. These videos can reveal useful stories, language, and objections. ### 10. Carry the same promise through the purchase process **Source — P2 18:41–19:07; P4 06:18–06:44; P8 19:50–20:31.** Joseph connects customer selection with the product, onboarding, screenshots, and paywall. **Action.** Compare the video promise with the store page and first app session. Remove changes in meaning. The person should find the outcome that the video led them to expect. ## B. Research formats that can produce customers ### 11. Separate attention from customer results **Source — P2 09:08–09:45.** Joseph rejects maximizing views without users or revenue. **Action.** Review views beside installs, activation, payments, and retained use. A popular video can fail as a customer source. A smaller video can reach more suitable customers. ### 12. Study a repeatable structure **Source — P2 09:45–10:59.** Joseph says an effective format can be adapted and repeated. The first success provides a starting point. **Action.** Describe the format as a sequence: opening statement, problem, evidence, product role, and next step. Save this structure with the source video. Do not record only its subject. ### 13. Check whether paid promotion explains the views **Source — P3 01:12–03:15.** Joseph shows a video with about 900,000 views and three comments. He identifies it as a promoted advertisement. **Action.** Check comments, reactions, account history, and any available promotion information. A low comment count is a reason to investigate. It does not prove that a video received paid promotion. ### 14. Paid success does not establish organic success **Source — P3 02:26–03:38; P8 12:15–12:55.** Joseph warns that a direct product demonstration may work as an advertisement but fail without paid reach. **Action.** Record whether each reference has evidence of organic distribution. Evaluate paid and organic results separately. A format can work in one setting and fail in another. ### 15. Test a person speaking directly to the camera **Source — P3 03:15–04:03; P5 37:30–38:12.** Joseph favors talking-head videos. He says trust helps these videos produce customers. **Action.** Test a credible person explaining a real problem and its solution. Compare the customer results with a short demonstration. Treat his preference as a test proposal for this app. ### 16. Make the product a natural part of the explanation **Source — P5 36:31–38:12.** A participant asks about a popular video with weak conversion. Joseph advises discussing problems and solutions within the relevant topic. **Action.** Write the explanation first. Introduce the app at the point where it helps solve the stated problem. Make the next action clear and relevant. ### 17. Borrow from adjacent subjects when direct examples are scarce **Source — P8 03:32–04:20.** Joseph discusses message-based products. Familiar message conversations may provide formats even when direct competitors have few examples. **Action.** Find familiar situations that show the product's value. Adapt their structure carefully. Record the adaptation as an unproven test, rather than a confirmed result. ### 18. Use tools to reduce research time **Source — P5 26:32–27:55; P8 07:09–07:43.** Joseph describes SpyTalk as a research tool. He says it does not remove the need to produce and test videos. **Action.** Measure research tools by the useful test references they produce. Avoid building another tool before using the evidence already available. ### 19. Use engagement thresholds as screening rules **Source — P8 09:24–10:10.** Joseph suggests references with at least 100,000 views and about 5% engagement. **Action.** Record the exact engagement formula. Apply the same formula across comparisons. These are his research rules, not universal platform standards. ### 20. Do not require a million views before continuing a test **Source — P6 01:35–02:26.** Joseph tells a participant to continue improving a format that reaches 30,000–50,000 views. **Action.** Examine customer quality and repeatability at moderate reach. Continue a useful format when the results justify another test. ## C. Select and manage creators ### 21. Begin with two to five creators **Source — P2 11:27–12:38; P8 05:08–05:51.** Joseph recommends at least two creators and no more than five initially. **Action.** Start with a group small enough for close review. Two creators provide a comparison. They do not, by themselves, create a controlled experiment. ### 22. Select creators for trust **Source — P3 04:27–05:39.** Joseph calls his two trust types the “Mountain and the Mirror.” One provides credible authority. The other shares the viewer's experience. **Action.** Decide which type fits each test. Select a person with genuine experience or qualifications. Do not write invented credentials into a script. ### 23. Review relevant examples before agreeing to higher pay **Source — P5 33:06–35:20.** Joseph recommends higher pay for creators with evidence of strong work, including successful app videos. **Action.** Ask for relevant examples and the creator's role in each one. Examine delivery, audience fit, and customer results where available. ### 24. Include response to feedback in the selection process **Source — P8 01:17–02:54.** Joseph says creators need to accept direction and be easy to work with. **Action.** Use a small paid trial. Give one clear revision request. Assess whether the creator understands and applies it. Attractive samples alone do not show this ability. ### 25. Separate production volume from recruitment volume **Source — P5 35:20–36:31.** Joseph describes contacting many candidates to find a small number of suitable creators. **Action.** Keep the active creator group small while maintaining a wider candidate list. Do not hire every available person merely to increase posting volume. ### 26. Use his pay examples as starting assumptions **Source — P5 33:06–35:20; P8 06:31–07:09.** Joseph gives a monthly base of approximately $300–$800 for daily posts. He gives bonus examples of $0.50–$2 per thousand views. **Action.** Obtain actual proposals for the required work. Confirm whether the price includes scripts, revisions, posting, platform versions, and usage rights. His figures are not verified market quotations. ### 27. Define the bonus before work starts **Source — P3 09:40–10:16; P5 33:50–35:20.** Joseph combines a base payment with performance incentives. **Action.** Define eligible views, the measurement period, excluded paid views, payment timing, and any cap. Record the approved accounts. A view bonus does not directly reward purchases. ### 28. Define deliverables in the agreement **Source — P5 38:12–39:01; P8 00:34–01:17.** Joseph discusses about 30 monthly videos. He says account preparation should be included in the agreed work. **Action.** State the number of videos, posting frequency, revision allowance, account preparation, and cancellation process. Confirm account access and video usage rights. ### 29. Give each creator a dedicated account **Source — P5 39:41–41:54; P6 02:26–03:13.** Joseph recommends a new account for each creator and app. He discourages several creators sharing one page. **Action.** Keep the audience and creator presentation consistent. Document account ownership and access from the start. His explanation of platform behavior remains unverified. ### 30. Separate account preparation from creative performance **Source — P5 41:12–41:54; P6 01:35–02:26.** Joseph discusses account preparation and restricted reach. He does not give a complete account preparation procedure. **Action.** Record account status and publishing conditions. Do not diagnose every weak video as an account restriction. Ask for the current written procedure when joining the program. ### 31. The founder does not need to perform every video **Source — P5 47:16–47:51.** Joseph treats delivery, editing, pacing, and text placement as skills. Founders can develop these skills or hire someone who has them. **Action.** Let the founder own the customer problem and test decisions. Assign performance and editing to the person who can execute them well. ### 32. Use marketplaces according to the current task **Source — P5 06:23–08:17; P6 03:13–04:19.** Joseph favors Sideshift for finding and managing creators. He presents Noise as more suitable for repeating an established format. **Action.** Distinguish finding a format from producing more of it. Judge candidates and tools against the current task. These are Joseph's opinions, not a current independent product comparison. ## D. Run tests that teach something ### 33. Give every video a test question **Source — P2 11:27–12:14.** Joseph describes each format as a hypothesis and each video as a test. **Action.** Write the expected result before publication. Example: “A story about a failed study routine will attract learners who need more structure.” Define what evidence would support it. ### 34. Change few variables at once **Source — P2 11:51–12:38.** Joseph recommends testing only a small number of variables at a time. **Action.** Change the opening sentence while keeping the creator and main explanation stable. Then test another creator. Large simultaneous changes make the result harder to explain. ### 35. Separate the creator from the format **Source — P3 05:39–06:31; P8 01:17–02:54.** Joseph says poor results can come from the format or the creator. **Action.** Let two creators test a similar structure. Let each creator also test another structure. This provides better evidence than replacing everyone after one weak result. ### 36. Review the parts of the video **Source — P3 05:39–06:31; P5 47:16–47:51.** Joseph names the opening, screen text, tone, delivery, pacing, and format as possible causes. **Action.** Review each element with a specific observation. Replace “make it better” with a clear change, such as a shorter opening or a more visible product result. ### 37. Seek review from people with practical experience **Source — P3 06:05–06:31.** Joseph values feedback from people who have made successful short videos. **Action.** Bring the actual video, its intended audience, and its results to the review. General marketing advice cannot explain a specific failure as well as direct examination. ### 38. Allow enough attempts to learn **Source — P4 04:10–05:00.** Joseph says the first strong result can require 60–80 videos. He says many people stop around 30. **Action.** Set a test budget and review schedule before starting. Continue only when the tests produce learning or promising results. Eighty repeated failures do not automatically establish progress. ### 39. Distinguish learning time from an unresponsive creator **Source — P8 01:17–02:54.** Joseph discusses ending a creator trial after roughly one or two weeks when there is no progress and poor response to feedback. **Action.** Assess reliability and response to direction separately from view counts. A responsive creator may need a better format. An unresponsive creator may require replacement. ### 40. Multiple formats can share one creator **Source — P8 11:34–12:15.** Joseph says an initial brief can contain several formats. **Action.** Give each format a name and record which one each video uses. This allows comparison without losing the creator's consistent presentation. ### 41. Use AI video as an early message test **Source — P5 29:44–30:24, 34:32–35:20; P8 07:43–11:34.** Joseph distinguishes testing an angle with AI from confirming a format with human creators. **Action.** Use a low-cost AI test to identify messages worth examining. Confirm customer response with credible human delivery. Do not equate AI video views with purchase intent. ### 42. Test before release, then connect people to the product **Source — P5 39:01–39:41, 46:39–47:16; P8 19:50–20:31.** Joseph supports marketing before launch. He also advises moving interested people into actual product use. **Action.** Provide a clear waitlist or test-access route. Record the promised benefit. Invite people to use the product when it is ready. Attention alone does not establish retention. ## E. Measure and expand useful results ### 43. The first month needs one main objective **Source — P3 06:31–06:57.** Joseph defines the first month's task as finding a working format. **Action.** Agree on a practical definition: the video reaches suitable people, produces useful customer action, and shows some repeatability. Do not require a particular revenue milestone without evidence. ### 44. A company account can attract attention without customers **Source — P3 06:57–08:42.** Joseph describes Lox Club's company account receiving many views with weak results. A separate account uses dinner-party content. **Action.** Audit what the account actually produces. Separate community entertainment from content that creates product demand. Both can have value, but they have different objectives. ### 45. Repeat successful structures **Source — P3 08:17–08:42; P5 40:31–41:12.** Joseph describes repeating a successful structure for months. **Action.** Produce several executions of a useful format. Keep the mechanism stable while changing a story, example, or opening. Monitor whether customer results remain useful. ### 46. Expand one variable at a time **Source — P3 10:16–11:14.** Joseph lists text openings, visual openings, creators, accounts, countries, and paid advertising as expansion variables. **Action.** Preserve a reference version. Compare each new version against it. Record which changes improve results and which only increase production. ### 47. Publish on the platforms he names **Source — P5 41:12–41:54.** Joseph says his TikTok discussion also includes Reels and Facebook. **Action.** Test suitable versions on these platforms when they fit the audience. Record results separately. Do not assume the same video has the same audience or performance everywhere. ### 48. Use customer economics before large expansion **Source — P3 09:07–10:16.** Joseph gives an example of 100,000 views, 500 downloads, and 10% payment conversion. Revenue figures are omitted. **Action.** Replace every assumption with observed app data before increasing spend. The example implies 50 paying customers. The transcript does not define the revenue period. ### 49. Correct the revenue-versus-profit distinction **Source — P3 09:40–10:16.** Joseph corrects himself from “profit” to revenue. He compares gross revenue with direct video cost. Revenue figures are omitted. **Action.** The comparison measures revenue relative to direct video cost. It is not verified net profit. Include unsuccessful videos, bonuses, fees, refunds, tools, and operating time in a full result. ### 50. Include failed tests in acquisition cost **Source — P3 10:16–10:52; P4 04:10–05:00.** The talk combines strong individual video economics with many attempts before success. **Action.** Divide the full test cost by the paying customers attributable to that test period. One successful video does not erase the cost of the preceding attempts. ### 51. Treat paid advertising as another test **Source — P4 00:10–00:35; P8 12:15–12:55, 20:31–21:12.** Joseph favors organic work first. He also allows small paid tests early and mentions $50 per day as an example. **Action.** Separate a small paid test from a decision to expand spending. Judge paid performance using actual customer economics. Engagement alone does not establish a profitable advertisement. ### 52. Influencer access and creator production are different purchases **Source — P8 15:13–16:36.** Joseph says influencers charge for access to an existing audience. A creator using a new account mainly supplies production skill. **Action.** Compare the cost of audience access with the cost of making repeatable content. Confirm which service the agreement buys. Use each service only when its measured result supports it. ## F. Make the system useful to the team ### 53. Reserve approximately five hours each week **Source — P5 18:25–19:34.** Joseph estimates about five weekly hours for the founder's work. He says this time may need to replace some coding time. **Action.** As an initial allocation, use one hour for customer research, one for briefs, one for creator review, and two for analysis and decisions. Adjust from actual use. ### 54. Use short, regular reviews **Source — P5 20:38–21:26, 23:51–24:38.** Joseph presents frequent reviews and community contact as ways to maintain progress. **Action.** Hold a weekly review with four outputs: what happened, what it suggests, what changes, and who completes the next test. ### 55. Evaluate the community through useful work **Source — P5 12:14–13:31, 31:08–33:06.** Joseph values complementary skills and active community leadership. He says a group chat alone is insufficient. **Action.** Find members who can examine current problems. Offer the team's own skills in return. Measure useful reviews and completed decisions, rather than message volume. ### 56. Bring evidence to the mastermind **Source — P4 08:04–11:50; P5 13:31–14:11.** Joseph describes an initial strategy call, course material, and three weekly coaching calls. **Action.** Prepare the app, audience hypotheses, current videos, costs, and customer results before the first call. Ask the coaches to choose the first tests with you. ### 57. Use included material to reduce delay **Source — P5 05:23–13:31.** The offer includes creator contacts, procedures, agreement templates, research access, and creator-management access. **Action.** Use these materials to start the work. Confirm which tasks each template covers. Avoid rewriting the whole operating process before the first creator trial. ### 58. Make each coaching call produce a decision **Source — P4 06:18–06:44; P5 20:38–21:26.** Joseph emphasizes implementation and diagnosis. **Action.** End each call with an approved audience, format, revision, or replacement decision. Record the owner and date. General encouragement does not substitute for a next test. ### 59. Preserve the skill after the membership ends **Source — P5 22:39–23:51.** Joseph presents this work as a reusable marketing skill. **Action.** Maintain a library of references, briefs, results, creator reviews, and successful structures. Record why a decision changed. Future work needs evidence, not only the final successful video. ### 60. Let customer response change the product **Source — P2 18:41–19:07; P5 39:01–39:41; P8 19:50–20:31.** Joseph links marketing tests to product positioning and onboarding. **Action.** Record repeated customer questions and unmet expectations. Use them to improve the product promise and first useful experience. More content cannot repair a promise that the product fails to meet. ## Proposed 30-day application This is a proposed work plan. Joseph does not prescribe this exact schedule. A successful format may require more than 30 days. ### Days 1–3: choose the problem - Select one app for the first coordinated test. - Write three customer hypotheses. - Describe the problem, present alternative, and expected result for each. - Select one primary hypothesis for initial work. - Check that the product can deliver the stated result. - Record current installs, activation, payments, and retention where available. ### Days 4–7: prepare references and creators - Collect 15–20 relevant video references. - Mark each reference's evidence of organic reach. - Identify three repeatable structures. - Select two creators for paid trials. - Prepare account ownership, deliverables, and bonus terms. - Write a brief with the audience, promise, structure, required evidence, and next action. ### Days 8–14: begin production - Start the agreed posting schedule. - Keep account preparation visible in the schedule. - Record the format and test question for every post. - Review delivery and reliability early. - Collect customer results with views. - Change one major element at a time where practical. ### Days 15–21: diagnose - Compare the same format across creators. - Compare different formats within each creator's work. - Review comments for customer intent and confusion. - Check whether the store page continues the video promise. - Bring the strongest and weakest examples to coaching. - End arrangements that show poor reliability or poor response to feedback. ### Days 22–30: confirm or revise - Repeat the strongest structure several times. - Confirm whether it produces useful customer action again. - Calculate the cost of the entire test period. - Increase production only when the evidence supports it. - Otherwise, revise the audience, promise, or format. - Set the next review date and maximum test spend. ## Example messages for a Chinese-learning app These are proposed hypotheses. They are not Joseph's examples. They do not assert existing product capabilities. | Possible audience | Problem to examine | Video direction | Product evidence needed | |---|---|---|---| | Learner preparing for an exam | Remembers isolated words but struggles with use | A learner explains a specific study failure | The app actually supports the tested exam need | | Adult with a Chinese-speaking partner | Cannot join ordinary family conversations | A personal story about a missed conversation | Relevant lessons or practice exist | | Traveler or recent arrival | Cannot complete a common daily task confidently | A short account of one difficult situation | The app supports that situation accurately | | Learner who repeatedly stops studying | Cannot maintain a useful routine | A creator explains a realistic study routine | The app supports the routine shown | Choose an audience only after checking product fit and actual audience evidence. Use genuine experience in personal stories. ## One record for each test | Field | Purpose | |---|---| | App and audience | Identify who should benefit | | Customer problem | State the reason to act | | Source reference | Preserve the original evidence | | Format and opening | Identify what the video tests | | Creator and account | Separate person and account effects | | Platform and publication date | Compare equivalent observation periods | | Views and engagement | Measure attention | | Clicks and installs | Measure movement toward the app | | Activation and payments | Measure useful customer action | | Revenue period | Prevent mixed financial comparisons | | Base cost and bonus | Include the actual production expense | | Attribution method | State how confidently results connect to the video | | Decision and next test | Turn the result into work | If direct attribution is unavailable, label the estimate. Do not assign all new customers to a video merely because publication dates overlap. ## Economics: apply the method without storing revenue figures Record views, download conversion, payment conversion, and the full production cost. Separate gross revenue from profit. Include unsuccessful tests, fees, refunds, tools, and operating time. Keep revenue figures outside the brain under company policy. The source example does not establish recurring revenue, net profit, or expected company results. At Joseph's stated base range, two creators cost about $600–$1,600 monthly before bonuses and other costs. Five cost about $1,500–$4,000. Actual agreements may differ. ## Prepare for joining the mastermind Since joining is likely, the useful next step is a concrete preparation package: 1. One app and one person responsible for the test program. 2. Three audience hypotheses with supporting references. 3. Existing videos, including poor results. 4. Current customer conversion data, with measurement limits. 5. A maximum monthly test budget. 6. Five reserved weekly hours. 7. The main decisions that need coaching. Ask the first strategy call to produce a selected audience, two or three formats, creator criteria, and a review date. Confirm the written terms for price, access, review availability, and the included tools. Joseph describes six months of mastermind access and twelve months of community access. He also describes three weekly calls, a three-hour course, and an initial strategy session. The transcript contains no clear final program price. The $15,000 monthly figure refers to his agency service, not the mastermind fee. The offer describes a creator list, three months of Sideshift, six months of SpyTalk, and agency procedures. Treat these as workshop offer statements until the written agreement confirms them. An additional AI tool receives discussion during questions; its delivery status is unclear. ## Source limits that affect decisions - The recording export contains severe repetition and occasional missing words. - Some app names, personal names, and revenue figures remain uncertain. - The time references are local to transcript parts. - Revenue stories illustrate Joseph's argument; they do not establish expected results for this team. - His claims about account preparation, restricted reach, and platform algorithms are not independently verified here. - The talk does not provide a complete attribution method or retention framework. - The talk does not give a complete creator contract or account preparation procedure. - Research attention, purchase conversion, and retained use remain separate forms of evidence. ## Source artifacts The original transcript, reading copy, and complete local report remain in the owner-provided local working folder. Raw transcript data is not copied into this shared page. Time references above preserve the source-part evidence. ## Related knowledge - [[people/joseph-choi]] - [[references/marketing/superwall-repeatable-app-growth-2026-09-23]] - [[playbooks/consumer-app-growth]] - [[playbooks/consumer-app-growth/promise-continuity]] This workshop record adds evidence and proposed tests. It does not replace existing app decisions or claim that a membership or campaign has started.