YouTube Analytics Lab
This project page turns the AugmentedThinker YouTube channel from a posting surface into a learning surface.
The goal is not to chase views as validation. The goal is to treat YouTube as an attention laboratory: publish small visual experiments, pull the available API data, identify what the signal suggests, and adjust the next experiment without overfitting to one result.
Current Snapshot
Queried 2026-06-08 at 12:51 UTC / 08:51 EDT.
Current Data API read: 1,645 channel views, 2,051 summed public video views, 21 total likes across public videos, and 0 comments. The Data API is the better current public count for newest uploads; the Analytics API is still lagging the newest June 6 and June 7 videos in day-level rows.
June 2 Baseline
Queried 2026-06-02 at 15:01 UTC / 11:01 EDT.
Laboratory frame: metrics are observations, not verdicts. A good video can underperform, and a weak video can overperform. The useful question is what each Short tested, what signal came back, and what small adjustment the next Short should make.
What People Are Looking For
Captured manually from YouTube Studio Analytics > Trends on 2026-06-02 around 11:35 EDT. These are the Studio recommendation queries from the screenshot, not Analytics API search-traffic rows.
Creative read: this is the better direction signal. It suggests the next Shorts should test practical OpenClaw use cases such as lead generation, personal assistance, marketing, tutorials, and podcast-style AI/robotics commentary. Treat this list as a backlog of prompts to test, not a guarantee that any one search will perform.
2026-06-08 Analysis 路 From first snapshot to daily loop Current read across public channel growth, top videos, the Red Pill signal, Analytics API lag, and the next behavior change.
Channel Snapshot
- Queried: 2026-06-08 at 12:51 UTC / 08:51 EDT.
- Channel: AugmentedThinker, handle
@augmentedthinker. - Channel ID:
UCHdJh8bMY8secEQeEBEbC1A. - Data API channel snapshot: 18 public videos, 6 subscribers, 1,645 channel views.
- Data API per-video public snapshot: 2,051 summed public video views, 21 likes, 0 comments.
- Analytics API May 25 through June 8 aggregate: 1,473 views, 15 likes, 0 comments, 121 estimated minutes watched, 16-second average view duration.
The Data API public counts are the most useful live read for the newest uploads. The Analytics API adds watch-time context, but its day-level rows are lagging the newest June 6 and June 7 videos, so do not treat its recent daily rows as complete.
Progress Since June 2
| Metric | June 2 Baseline | June 8 Current | Change |
|---|---|---|---|
| Public videos | 9 | 18 | +9 |
| Subscribers | 6 | 6 | 0 |
| Channel views | 1,057 | 1,645 | +588 |
| Summed public video views | 1,231 | 2,051 | +820 |
| Analytics API views | 1,069 | 1,473 | +404 |
| Estimated minutes watched | 93 | 121 | +28 |
Top Public Videos By Current Data API Views
| Published | Video | Length | Views | Likes | Comments |
|---|---|---|---|---|---|
| 2026-05-26 | OpenClaw Field Note 002: Alpine Signal Log #Shorts | 11s | 466 | 4 | 0 |
| 2026-05-25 | OpenClaw Robot Intro Short #Shorts | 25s | 374 | 4 | 0 |
| 2026-06-07 | OpenClaw Offers the Red Pill #Shorts | 11s | 344 | 5 | 0 |
| 2026-06-04 | OpenClaw Keeps the Loop Alive #Shorts | 24s | 196 | 2 | 0 |
| 2026-05-28 | OpenClaw Agent 004 Greeting #openclaw #aiagents | 26s | 165 | 0 | 0 |
| 2026-06-06 | Bounded Signals Become Action #Shorts | 24s | 134 | 1 | 0 |
Current Read
The channel has doubled its public video count since June 2, moving from 9 to 18 public videos. Subscribers are unchanged at 6, so the growth is still passive reach rather than community conversion. Channel views increased from 1,057 to 1,645, and the summed public video view count increased from 1,231 to 2,051.
The strongest new signal is OpenClaw Offers the Red Pill #Shorts: 344 views, 5 likes, 0 comments, and now the third-highest public video by Data API view count. It is only 11 seconds, conceptually simple, and visually centered on OpenClaw as a cinematic choice-giver. That supports the hypothesis that short, instantly readable OpenClaw identity concepts may outperform more internal workflow-language videos.
The daily cron loop is working as an operating system, but the best-performing recent upload came from a manual AI-generated movie. The next behavior change should be to let the daily routine continue, while using manual experiments to discover stronger visual and hook formats that can later influence the routine.
Updated Hypotheses
- Simple cinematic OpenClaw identity hooks may travel better than abstract signal-loop doctrine.
- Very short videos around 11 seconds can compete with or beat longer 24-second field notes when the premise is instantly legible.
- The daily cron is valuable as cadence and learning infrastructure, but manual AI-movie experiments may be the better style-discovery surface.
- Views are increasing, but subscriber conversion and comments are still flat. The channel has attention, not community yet.
- The next critique should compare daily-cron output against the Red Pill style signal, not only against previous cron-made field notes.
2026-06-02 Analysis 路 From channel ignition to first API snapshot Baseline read across channel stats, video stats, Analytics API rows, hypotheses, and next measurement fields.
API Boundary
The June 2 query used the local YouTube OAuth token and two API surfaces. The YouTube Data API returned channel metadata, the uploads playlist, video status, duration, processing, and public statistics. The YouTube Analytics API returned channel-level and older-video analytics for views, likes, comments, estimated minutes watched, and average view duration.
The Analytics API currently lags the newest June 1 uploads in the per-video report rows, so this first analysis combines current Data API public stats with Analytics API aggregate and older-video rows.
Traffic Source Snapshot
Christopher asked whether OpenClaw can access the YouTube Studio Analytics "Trends" style search suggestions that show what people are looking for. Current answer: partial access.
OpenClaw can query YouTube Analytics API traffic-source reports for YouTube Search terms that actually generated views for a specific video. This is not the same as the broader YouTube Studio Trends/Research panel that suggests related searches for future content direction.
Google's Analytics API documentation describes YT_SEARCH as YouTube search-result traffic and says insightTrafficSourceDetail can specify the search term when filtering for that source. The working query also requires a specific video filter, maxResults, and sorting. See Google's Analytics dimensions documentation and sample API requests.
| Traffic Source | Meaning | Views | Minutes Watched | Avg View Duration |
|---|---|---|---|---|
SHORTS | Vertical Shorts feed swipes | 950 | 62 | 13s |
YT_CHANNEL | Channel page | 57 | 21 | 30s |
YT_SEARCH | YouTube search results | 19 | 1 | 13s |
EXT_URL | External links and web referrals | 18 | 3 | 23s |
SUBSCRIBER | Subscription/homepage surfaces | 9 | 1 | 17s |
YT_OTHER_PAGE | Other YouTube pages | 7 | 0 | 7s |
Available Search Terms
The video-level YouTube Search terms available so far were tiny and not strategically useful yet:
| Video | Search Term | Views |
|---|---|---|
| OpenClaw Field Note 002: Alpine Signal Log #Shorts | multfilm uzbek tilida | 1 |
| OpenClaw Field Note 002: Alpine Signal Log #Shorts | multik | 1 |
| OpenClaw Robot Intro Short #Shorts | drunk magnus carlsen | 1 |
| OpenClaw Agent 004 Greeting #openclaw #aiagents | flugel re zero | 1 |
| OpenClaw Agent 004 Greeting #openclaw #aiagents | sesko jump celebration | 1 |
Search read: the channel is not yet getting meaningful search-led traffic. Most early attention is coming through the Shorts feed, not search. These terms should not guide the creative direction yet. If search terms begin showing repeated AI-agent, robot, automation, OpenClaw, or workflow-related queries, then those terms should influence hooks, titles, captions, and follow-up Shorts.
Channel Snapshot
- Queried: 2026-06-02 at 15:01 UTC / 11:01 EDT.
- Channel: AugmentedThinker, handle
@augmentedthinker. - Channel ID:
UCHdJh8bMY8secEQeEBEbC1A. - Channel created: 2026-04-26.
- Data API channel snapshot: 9 public videos, 6 subscribers, 1,057 channel views.
- Data API per-video current snapshot: 1,231 summed public video views.
- Analytics API May 25 through June 2 aggregate: 1,069 views, 10 likes, 0 comments, 93 estimated minutes watched, 16-second average view duration.
The totals do not reconcile perfectly because YouTube Data API public statistics and YouTube Analytics reports update on different schedules and use different aggregation rules. Treat them as directional signal, not a precise accounting ledger.
Current Public Video Stats
| Published | Video | Length | Views | Likes | Dislikes | Comments |
|---|---|---|---|---|---|---|
| 2026-05-25 | OpenClaw Field Note 001: Hello from the Workshop | 11s | 29 | 1 | 0 | 0 |
| 2026-05-25 | OpenClaw Robot Intro Short #Shorts | 25s | 372 | 4 | 1 | 0 |
| 2026-05-26 | OpenClaw Field Note 002: Alpine Signal Log #Shorts | 11s | 464 | 4 | 1 | 0 |
| 2026-05-26 | OpenClaw 05-26 #openclaw #aiagents #scifi | 24s | 26 | 0 | 0 | 0 |
| 2026-05-28 | OpenClaw Agent 004 Greeting #openclaw #aiagents | 26s | 165 | 0 | 0 | 0 |
| 2026-05-29 | AI AGENT VS. THE DEEP: OpenClaw Robot | 29s | 20 | 1 | 0 | 0 |
| 2026-06-01 | Introduction to OpenClaw AI Persona | 19s | 35 | 1 | 0 | 0 |
| 2026-06-01 | OpenClaw Pipeline Wakes #Shorts | 21s | 62 | 1 | 0 | 0 |
| 2026-06-01 | Automated Content Creation with an AI Robot | 21s | 82 | 1 | 1 | 0 |
Analytics API By Day
| Day | Views | Likes | Comments | Minutes Watched | Average View Duration |
|---|---|---|---|---|---|
| 2026-05-25 | 336 | 5 | 0 | 34 | 20s |
| 2026-05-26 | 288 | 1 | 0 | 23 | 14s |
| 2026-05-27 | 236 | 3 | 0 | 21 | 11s |
| 2026-05-28 | 179 | 0 | 0 | 10 | 27s |
| 2026-05-29 | 28 | 1 | 0 | 3 | 16s |
| 2026-05-30 | 2 | 0 | 0 | 0 | 0s |
First Read
The early channel has real signal for such a young experiment. Two videos crossed 300 views, one crossed 400, and multiple later videos reached tens of views within the first day. Comments are still zero, so the current signal is mostly passive attention rather than conversation.
The highest-view videos suggest that compact OpenClaw robot identity pieces and field-note Shorts are the strongest early shape. The newest June 1 videos are still too fresh for Analytics API rows, but Data API stats already show they are being seen.
First Hypotheses
- Shorter OpenClaw field-note videos may travel better than longer narrative descriptions.
- A recognizable robot/persona visual identity matters more than detailed explanatory metadata at this stage.
- Videos that feel like a world someone is entering may be more promising than videos that only describe a workflow.
- Pipeline/process videos are strategically important even if they are not the highest-reach format, because they teach the production loop and can become future digital-product material.
- The channel needs postmortem discipline before it needs more automation.
Next Measurement Template
- Video ID and URL.
- Title and published date/time.
- Concept category, hook type, visual style, and length.
- Hypothesis before posting.
- Data API views, likes, dislikes, and comments after 24 and 72 hours.
- Analytics API retention/watch-time metrics once available.
- Qualitative read and one next adjustment.
Next action: create the next Short with one explicit hypothesis before upload, then return here after 24-72 hours for a second collapsed analysis drawer.