簡易檢索 / 詳目顯示
| 研究生： |
李約希 Lee, Yueh-Hsi |
|---|---|
| 論文名稱： |
LOOM：AI生成之遊戲化知識學習平台商業模式研究 LOOM: A Business Model Study of an AI-Generated Gamified Knowledge Learning Platform |
| 指導教授： |
何乾瑋
Ho, Chien-Wei |
| 口試委員: |
白佩玉
曾忠蕙 |
| 學位類別： |
碩士
Master |
| 系所名稱： |
商學院 - 國際經營管理英語碩士學位學程(IMBA) International MBA Program College of Commerce(IMBA) |
| 論文出版年： | 2026 |
| 畢業學年度： | 114 |
| 語文別： | 英文 |
| 論文頁數： | 87 |
| 中文關鍵詞： | 遊戲化學習 、微學習 、行動學習 、商業模式 、使用者留存 |
| 外文關鍵詞： | Gamified learning, Microlearning, Mobile learning, Business model, User retention |
| 相關次數： | 點閱：11 下載：0 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
本論文為 LOOM 的商業計畫研究。LOOM 是一個以人工智慧生成內容驅動的遊戲化微學習平台。研究的起點是教育科技市場的一個結構性錯配:多數學習產品是為三十到六十分鐘的學習時段所設計,但現代使用者卻是以兩分鐘的碎片時間在學習。這個錯配反映在三個產業特有的數據上:學習應用程式單次使用時間的中位數不到三分鐘、線上課程完成率僅五到十五個百分比,以及超過八成的使用者在安裝後三十天內流失。本論文將此一現象視為「設計假設」的問題,而非「手機習慣」的問題,並探討:若一個學習產品從一開始就接受「碎片化注意力」為常態,它會是什麼樣貌。
本論文提出的答案即為 LOOM。使用者一次的學習歷程為五道題目、費時一到三分鐘;每一題作答時間十到十五秒,結束時立即提供回饋與簡短解說,並累積至一項自訂的參與指標——智慧指數(Wisdom Index)。智慧指數與每日連續紀錄(streak)相互搭配:連續紀錄提供短期的習慣壓力,智慧指數則累積長期的學習進展,且不會因連續紀錄中斷而歸零。人工智慧生成內容讓平台能以數週、而非數月的速度推出新科目,使單一產品得以涵蓋使用者目前分散於不同應用程式的多種學習主題。
本商業模式以「可行性」為目標,而非以創投規模的成長性為訴求。在保守假設下,並以新台幣八百萬元(約二十五萬美元)、且不經創投募資的起始資本推動,LOOM 可於第四年達到 EBITDA 損益兩平,並於第五年達到約一百三十萬美元的營收。本論文以財務可行性——亦即一家小而專注、能靠自身資金存活的公司——作為衡量成功的標準。
This thesis presents a business plan for LOOM, a gamified micro-learning platform powered by AI-generated content. The starting point is a structural mismatch in the EdTech market: most learning products were designed for thirty- to sixty-minute study sessions, while modern users learn in two-minute fragments. The mismatch shows up in three category-specific numbers: a median learning-app session under three minutes, course completion of five to fifteen percent, and more than eighty percent of users gone within thirty days of installing. This thesis treats the mismatch as a design-assumption problem rather than a phone-habit problem, and asks what a learning product would look like if it accepted micro-attention as its default condition.
The proposed answer is LOOM. A LOOM session is five questions in one to three minutes. Each question runs for ten to fifteen seconds, ends in instant feedback and a short explanation, and contributes to a proprietary engagement metric called the Wisdom Index. The Wisdom Index is paired with a daily streak: the streak supplies short-term habit pressure, while the Wisdom Index accumulates a long-term record of progress that does not reset when the streak breaks. AI-generated content lets the platform launch new subjects in weeks rather than months, so a single product can cover the many subjects users currently scatter across separate apps.
The business model is built for feasibility rather than venture-scale attractiveness. On conservative assumptions, and with starting capital of NT$8 million (about USD 250,000) raised without a venture round, LOOM reaches EBITDA breakeven in Year 4 and roughly USD 1.3 million of revenue in Year 5. The thesis treats financial feasibility, meaning a small, focused company that can survive its own runway, as the standard for success.
Introduction 7
Personal Motivation 8
Research Approach 9
Thesis Structure 10
I. The Pain — Three Concrete Frustrations 11
A. The Time Problem 12
B. The Stickiness Problem 13
C. The Compounding Problem 14
D. The Learning Science Behind LOOM 14
II. The Market — From EdTech to LOOM’s Specific Lane 19
A. Global EdTech Context 19
B. TAM, SAM, and SOM 20
C. Why Asia First 21
D. Who Hurts Most — Reading the Market by Need, Not Only by Geography 22
III. The LOOM Platform 25
A. Core Design — A Ten-Second Question Loop 25
B. How a Session Works 25
C. AI-Generated Content Engine 26
D. Pain-to-Solution Mapping 28
IV. Stickiness — Daily Streak and Wisdom Index 29
A. The Two-Layer Engagement Design 29
B. Wisdom Index Scoring Mechanics 30
C. Decay Rates and Tier Thresholds 31
D. Why the Two-Layer Design Works 32
V. Competitive Landscape 34
A. Duolingo 35
B. Blinkist 35
C. Brilliant 36
D. Candy Crush — The Attention Reference 36
E. Each Competitor’s Specific Gap 38
F. SWOT Summary 39
VI. Defensibility — Why It Is Hard to Copy 42
A. Speed Advantage — AI-Native Content Engine (Y1) 42
B. Data Moat — Personalized Difficulty (Y2 onward) 43
C. Localization Moat — Asia-First Content (Y3 onward) 43
D. Why Duolingo Cannot Just Copy This 44
VII. Target Users — Segmented by Motivation 46
A. The Micro-Moment Learner 48
B. The Test Prepper 48
C. The Curious Professional 49
D. Enterprise Users (Deferred) 50
VIII. Business Model — Short, Mid, and Long Term 51
A. Year 1: Freemium and Light IAP 51
B. Years 2–3: Premium Subscription and Certifications 52
C. Year 4 and Beyond: B2B and Creator Marketplace 53
IX. Product Development — Short, Mid, and Long Term 54
A. Year 1: Minimum Viable Product 54
B. Years 2–3: Premium Tier and Certifications 55
C. Year 4 and Beyond: Platform and Ecosystem 56
D. Content Engine and Quality Assurance 56
X. Go-to-Market — Short, Mid, and Long Term 61
A. Year 1: Taiwan with Campus Program 61
B. Years 2–3: Japan, Korea, Southeast Asia 62
C. Year 4 and Beyond: English-Speaking Markets and B2B 63
XI. Financial Outlook and Feasibility 65
A. Six Core Assumptions 65
B. Five-Year Detail (Base Case) 66
C. Three Scenarios — Bear, Base, Bull 67
D. Sensitivity Analysis 69
E. Funding Structure 70
XII. Team — Why This Two-Person Setup Works 72
A. Founder 72
B. Engineering Co-Founder 72
C. Advisor 73
D. The AI-Native Advantage 73
XIII. Risk Analysis — Five Material Risks 75
A. AI Content Errors 76
B. Competitive Entry 76
C. Privacy and Regulation 77
D. Engagement Decay 77
E. Running Out of Money 78
XIV. Conclusion 79
A. Limitations and Future Research 80
References 83
Appendix — Detailed Financial Assumptions 86
A. OpEx Decomposition (Base Case) 86
B. Sensitivity Ranges 86
Activision Blizzard, Inc., 2023, Annual Report on Form 10-K for the Fiscal Year Ended December 31, 2022 (United States Securities and Exchange Commission).
AppsFlyer, 2024, Mobile app retention and engagement benchmarks by category, retrieved from AppsFlyer’s published market data.
Blinkist, 2024, Company overview and user statistics, retrieved from the company’s investor and product communications.
Brilliant.org, 2024, About Brilliant: User and product overview, retrieved from the company’s public communications.
Cepeda, Nicholas J., Harold Pashler, Edward Vul, John T. Wixted, and Doug Rohrer, 2006, Distributed practice in verbal recall tasks: A review and quantitative synthesis, Psychological Bulletin 132, 354–380.
CFA Institute, 2025, CFA Program exam results, February to November 2024 windows, retrieved from CFA Institute press communications.
Chun Shin Limited, various years, TOEIC Listening and Reading Test: Taiwan-area annual test-taker statistics, retrieved from the TOEIC Taiwan official website.
Chun Shin Limited, 2021, Taiwan Mid-to-Large Enterprise and Job-Seeker Foreign-Language Competency Management Survey (Chun Shin Limited, Taipei).
Csikszentmihalyi, Mihaly, 1990, Flow: The Psychology of Optimal Experience (Harper & Row, New York).
DataReportal, 2025, Digital 2025 Global Overview Report (DataReportal and Kepios, Singapore).
Deci, Edward L., and Richard M. Ryan, 2000, The “what” and “why” of goal pursuits: Human needs and the self-determination of behavior, Psychological Inquiry 11, 227–268.
Duolingo, Inc., 2021, Form S-1 Registration Statement (United States Securities and Exchange Commission).
Duolingo, Inc., 2026, Annual Report 2025, Form 10-K (United States Securities and Exchange Commission).
Fortune Business Insights, 2025, Education Technology (EdTech) Market Size, Share and Industry Analysis (Fortune Business Insights, Pune).
Hamari, Juho, Jonna Koivisto, and Harri Sarsa, 2014, Does gamification work? A literature review of empirical studies on gamification, in Proceedings of the 47th Hawaii International Conference on System Sciences, 3025–3034.
HolonIQ, 2025, Global education technology market size and outlook, retrieved from HolonIQ’s published market intelligence.
Japan Foundation and Japan Educational Exchanges and Services, 2025, JLPT statistics: Numbers of applicants and examinees, retrieved from the official JLPT website.
Ji, Ziwei, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Ye Jin Bang, Andrea Madotto, and Pascale Fung, 2023, Survey of hallucination in natural language generation, ACM Computing Surveys 55, 1–38.
Khan Academy, 2024, Khanmigo: AI tutor product communications and accuracy reporting, retrieved from Khan Academy public communications on its AI tutor.
Kivetz, Ran, Oleg Urminsky, and Yuhuang Zheng, 2006, The goal-gradient hypothesis resurrected: Purchase acceleration, illusionary goal progress, and customer retention, Journal of Marketing Research 43, 39–58.
Kumar, Vineet, 2014, Making “freemium” work, Harvard Business Review 92, 27–29.
Lally, Phillippa, Cornelia H. M. van Jaarsveld, Henry W. W. Potts, and Jane Wardle, 2010, How are habits formed: Modelling habit formation in the real world, European Journal of Social Psychology 40, 998–1009.
Landers, Richard N., Kristina N. Bauer, and Rachel C. Callan, 2017, Gamification of task performance with leaderboards: A goal setting experiment, Computers in Human Behavior 71, 508–515.
Mark, Gloria, Daniela Gudith, and Ulrich Klocke, 2008, The cost of interrupted work: More speed and stress, in Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 107–110.
Moore, Geoffrey A., 1991, Crossing the Chasm: Marketing and Selling High-Tech Products to Mainstream Customers (HarperBusiness, New York).
Reich, Justin, and José A. Ruipérez-Valiente, 2019, The MOOC pivot, Science 363, 130–131.
RevenueCat, 2024, State of Subscription Apps 2024 (RevenueCat, Inc., San Francisco).
Ries, Eric, 2011, The Lean Startup: How Today’s Entrepreneurs Use Continuous Innovation to Create Radically Successful Businesses (Crown Business, New York).
Roediger, Henry L., III, and Jeffrey D. Karpicke, 2006, Test-enhanced learning: Taking memory tests improves long-term retention, Psychological Science 17, 249–255.
Rowland, Christopher A., 2014, The effect of testing versus restudy on retention: A meta-analytic review of the testing effect, Psychological Bulletin 140, 1432–1463.
Sailer, Michael, and Lisa Homner, 2020, The gamification of learning: A meta-analysis, Educational Psychology Review 32, 77–112.
Sensor Tower, 2024, Mobile app engagement and session-length benchmarks, retrieved from Sensor Tower’s published market intelligence.
Slamecka, Norman J., and Peter Graf, 1978, The generation effect: Delineation of a phenomenon, Journal of Experimental Psychology: Human Learning and Memory 4, 592–604.
Statista, 2025, Mobile internet usage worldwide, retrieved from the Statista statistics portal (Statista, Hamburg).
