RnD · Линия доказательств Global validation
Каталог и оглавление
GVL.06 RnD/research/global-validation/06-source-ledger.md raw.md ->

06. Реестр источников

Дата проверки: 23 июля 2026 года.

Классы:

  • A — официальный первичный источник или academic paper;
  • B — источник с раскрытой методикой / marketplace listing / platform case study;
  • C — modelled traffic/revenue, company self-report без аудита или единичный signal.

Комбинированная маркировка вроде A/C означает официальный первичный URL, но содержащийся в нём scale/performance counter остаётся неаудированным self-report; B/C — раскрытый case-study контекст с тем же ограничением результата.

Рынок и аудитория#

ID Класс Источник Использование и ограничение
GV001 A/C D&D Beyond — Thanking Our Community for a Great 2024 85M fans, 19M users, 1.3M maps. First-party category counters, не paying groups
GV002 A/C Roll20 15M+ players, 1,200+ systems. First-party platform claim, не MAU
GV003 A/C StartPlaying — GMs earned over $17M $17M+ GM payout и 50K+ players к Dec 2023. First-party marketplace totals
GV004 B One-shot $10, $15, $21 Три проверяемых live listing prices/durations; offer, не доказанная продажа или market average
GV005 B Compare the Market / PureProfile Tabletop Gamers Survey Jan 2025, US n=1,013, CA n=1,010, AU n=1,012; mixed TTRPG+wargames+collecting question
GV006 B/C Discord case study — Death by AI / Playroom Nearly 7M users; in the first week >70% of sessions had more than three friends; >30% D7 engagement and thousands of first-month subscribers. Vendor/customer case, adjacent party game
GV007 A/C Latitude press AI Dungeon first-party scale/profitability and Voyage launch; not audited
GV008 A Kickstarter — 2024 Games/tabletop results Category creator spend, не SaaS revenue
GV009 A CALYPSO: LLMs as Dungeon Masters' Assistants AI supports prep/improv while human controls table; research prototype, не commercial PMF
GV011 A Co-Creativity at the Table LLM strengths in ideas/description and weaknesses in coherence/social GM work; 2026 preprint
GV012 A Google Trends — FAQ about data Relative sampled 0–100 data and interpretation limits
GV013 B SerpApi Google Trends API Retrieval intermediary for Google Trends UI-normalized time series, geography and related queries; not the invite-only official Google Trends API alpha

Конкуренты#

ID Класс Источник Использование и ограничение
GV010 B/C Local competitor vitality dossier, review synthesis, traffic/revenue/acquisition, source ledger 3,012-review corpus, product census, modelled traffic/revenue, acquisition evidence. Каждая оценка наследует ограничения своего первичного источника

Инфраструктура и платежи#

ID Класс Источник Публичный тариф/роль
GV020 A OpenAI API pricing GPT-5.4 mini $0.75/M input, $0.075/M cached, $4.50/M output на дату проверки
GV021 A OpenAI TTS-1 $15/M characters
GV022 A OpenAI image generation pricing GPT Image 1 Mini 1024² $0.005/$0.011/$0.036 low/medium/high; selected calculator also prices text prompt input at $2/M tokens
GV023 A OpenAI omni-moderation Text/image moderation free; audio requires STT; human review не бесплатен
GV024 A Daily Video SDK pricing Audio-only $0.00099/participant-minute after first 10K; realtime transcription $0.0059/unmuted participant-minute
GV025 A LiveKit pricing Ship $50/month, included WebRTC/egress and metered overage on date checked
GV026 A Google generative AI pricing Gemini 2.5 Flash Standard, explicit cache storage, Gemini 2.5 Flash TTS and Imagen 4 Fast rates used in Path B
GV027 A Google Speech-to-Text, Text-to-Speech STT V2 Standard $0.016/min; TTS rates per selected model
GV028 A Stripe pricing US domestic 2.9% + $0.30; international/FX/tax add-ons vary
GV029 A Paddle pricing Merchant of Record 5% + $0.50 public price
GV030 A Apple Small Business Program 15% qualifying commission
GV031 A Apple standard commission schedule Standard schedule/context; verify latest country contract before launch
GV032 A Google Play service fees Country/program/install-date-dependent structure; no single global fee
GV033 A Cloudflare R2 pricing $0.015/GB-month Standard, operations, free direct egress; DB/compute excluded

Local evidence artifacts#

ID Артефакт Роль
L-GT-01 Non-RU Google Trends query spec Pre-registered terms/geographies/periods
L-GT-02 Historical Google Trends run manifest Timestamp, requests, failures and hashes from the original blocked run
L-GT-03 Historical Google Trends error ledger 39 HTTP 429 attempts; no quantitative Trends result in that run
L-GT-04 Historical US-only retry manifest, errors Clean geo=US bootstrap also failed with three HTTP 429 responses before any widget request
L-GT-05 Successful SerpApi manifest 19/19 terminal jobs, source-safe request parameters, timestamps and 49 artifact hashes
L-GT-06 Worldwide category summary, problem summary, US summary Derived relative indexes, coverage and within-request recent direction
L-GT-07 Category related queries, problem related queries Top/rising intent vocabulary and semantic-contamination audit
L-GT-08 Category geography, problem geography Within-location term composition; not absolute country demand
L-ECON-01 Load scenarios CSV Explicit low/base/high assumptions and two vendor-path outputs
L-ECON-02 Free-trial sensitivity CSV Conversion × repeat sensitivity; not forecast
L-ECON-03 Component calculator CSV Selected model/rate, component cost and exact subtotal for every path × scenario

Правила использования#

  1. Company self-report помечается first-party и не превращается в audited fact.
  2. Similarweb/Semrush/Sensor Tower — modelled estimates, не MAU/revenue statements.
  3. Listing price — offer, не sale.
  4. Category audience не равна addressable party-first cohort.
  5. Все отсутствующие conversion, repeat, CAC, support и partner variables остаются unknown до pilot telemetry.