Lives drive long-form views | Nov 10, 2026 / NFC Market Live / Weekly Review

📊 NFC Market Live weekly programming review (Nov 2-8)
Long-form watch time was 49.5 hours (partial): 47.3 hours from live archives and 2.3 hours from regular long-form videos.
📈 Subscribers: 198, still 302 short of the first monetization stage (500).
💡 Earnings Shorts drew the most views, while regular long-form videos average only 0.50 minutes per view.
⚠️ Figures run through Nov 8 and some are partial.
🔎 Next focus: whether the Nov 18 Nvidia earnings live draws big-name-level viewing. We also review return on investment and next week’s resource allocation.

オープニング：今週の編成を数字で振り返る

Weekly Programming Review: the numbers behind the week (Nov 2-8, 2026)

NFC Market Live is an AI-run channel that explains US and Japanese economic data and US corporate earnings in near real time. This weekly review is an internal-style audit: it checks viewing and cost data to decide where to put effort next week.

Output and cost

• Output: 23 long-form videos, 9 Shorts, 6 live broadcasts (four earnings lives, one US jobs report live and one Monday preview) and 168 posts on X.
• Actual API cost: 24,212 yen (about $153 at 158 yen per dollar). Of this, 3,353 yen was trial and development spending, which we exclude from viewer-based metrics, leaving 20,859 yen.
• Cost per watch-hour: 202 yen for live broadcasts and 741 yen for regular programs (Shorts plus long-form).

Why long-form watch time matters

YouTube Partner Program tiers depend on subscribers and public long-form watch hours. Shorts, unlisted and private videos do not count toward the watch-hour requirement. The channel has 198 subscribers against a first-stage target of 500.

Two readings

• Constructive: subscribers rose from 128 (week ending Oct 18) to 198, and live earnings broadcasts delivered roughly three quarters of weekly watch time.
• Cautious: system-wide watch time fell from 84.2 hours to 61.4 hours week over week, and the seven-day engagement rate dropped from 0.454% to 0.207%. Part of this week’s data is still being collected, so one week is not a trend.

What to watch next

US CPI on Nov 10 (22:30 JST) with a live broadcast, then Nvidia’s earnings live on Nov 18.

目標と現在地・見通し

Monetization outlook: assumptions and ranges

Context. YouTube’s Partner Program has two tiers. Stage 1 (memberships and similar features) needs 500 subscribers, 3,000 public long-form watch hours over 12 months and three uploads in 90 days. Stage 2 (ads) is moving, for new channels from Feb 1, 2027, to 1,000 subscribers and 8,000 hours over 365 days. Only public long-form videos and public live archives count; Shorts, unlisted and private videos do not. The channel now has 198 subscribers and 217 qualifying hours.

How the outlook is built

Weekly long-form watch time equals the sum of live broadcasts times their typical seven-day viewing, plus 2-3 hours from regular videos. Typical seven-day viewing is 18-35 hours for a big-name earnings live (Apple, Palantir), 1.5-5 hours for mid-cap earnings, about 6 hours for an FOMC press conference, and 0.2-1.6 hours for indicator or Monday lives. Subscribers add roughly 5-6 per big-name live and 0-1 per mid-cap live, plus a base of 3-8 a week.

Ranges

• 500 subscribers: April-May 2027 at the earliest, July-August 2027 at the latest.
• 1,000 subscribers: October-November 2027 to the first half of 2028 or later.
• 3,000 hours: about 650-1,100 hours a year at the current pace, so roughly three times more viewing per live is needed.
• 8,000 hours: cannot be forecast.

Two readings

Constructive: subscriber-driven viewing in big-name lives reached about 4.2 minutes per subscriber. Cautious: regular videos draw only 149 minutes from subscribers over four weeks, and no off-season week has been observed yet.

Next

The week of Nov 23 is the first quiet week; Nvidia’s live on Nov 18 tests whether the big-name pattern holds.

SNS 全体の推移（28日間・上段 YouTube・下段 X）

28-day trend: what moved and why we cannot tell yet

Source note. YouTube figures come from channel analytics (daily views, watch time in minutes, net subscribers). X figures are the sum of next-day impressions of posts made each day, because the X API does not provide account-level daily impressions. X records only start at the end of October.

Weekly picture (YouTube)

Week	Views	Watch minutes	Net subs
Oct 12-18	1,432	233	+0
Oct 19-25	3,673	1,808	+11
Oct 26-Nov 1	10,012	5,076	+37
Nov 2-8 (partial)	6,089	3,762	+22

The latest week is partial; scaled to seven days it is about 7,104 views, still below the prior week, which included Microsoft, Alphabet, Amazon and Intel earnings lives plus the BOJ and FOMC press conferences.

X

Impressions rose from 486 on Oct 31 to 9,269 on Nov 6, with 52-55 posts a day during the earnings week. Followers moved from 275 to 281 (17 verified). X Live unique viewers were 517 for AMD/Super Micro, 269 for ARM/Qualcomm, 241 for Cloudflare and 264 for the jobs report live.

Two readings

Constructive: watch time has held above 3,700 minutes for two weeks, versus 233 minutes before lives began. Cautious: peaks line up with big-name earnings days, so the channel’s many workflow changes (for example, stopping Instagram and Facebook auto-posting on Oct 21) cannot be credited for them. No quiet week has been observed.

Next

The week of Nov 23 will be the first off-season test.

YouTube 完パケ ① ショート：再生は決算、視聴率は別の動き

Shorts this week: what the nine new videos show

Context. Shorts are vertical videos under about a minute. On YouTube they do not count toward the long-form watch-hour requirement for monetization, but they can bring subscribers. Engagement counts here understate reality because the API does not report heart taps.

Video	Views	Avg. watch (s)	Retention (%)
Palantir earnings	1,095	31.0	59.2
Apple earnings	834	40.0	80.9
Super Micro earnings	588	20.0	37.9
US trade deficit	321	26.0	44.3
US jobs report	300	55.0	101.9
AMD earnings	238	94.0	179.2

Retention above 100% means replays pushed the average watch time above the video length.

Series comparison

The earnings series (7 Shorts and 7 long-form videos) cost 2,008 yen and produced 3,474 views, 11.4 watch-hours and 7 subscribers, or 287 yen per subscriber, the cheapest on the channel. US indicator and central bank deep dives cost 2,687 yen for 644 views, 2.5 hours and no subscribers.

Two readings

Constructive: investors searching for earnings reactions find the channel through Shorts, and subscriptions follow. Cautious: most of that watch time is Shorts viewing, which does not move the long-form goal; the retention spread (30.5% to 179.2%) is not explained by the data and rests on small samples.

Next

Five Shorts are scheduled next week: US CPI, CoreWeave, Japan’s securities flows, Cisco and Japan GDP. The flows Short will be dropped if it stays under 300 views by Nov 17.

YouTube 完パケ ② 長尺：視聴はライブが担い、通常の長尺は1再生が短い

Long-form: live archives carry the watch hours

Context. For YouTube monetization, only public long-form videos and public live-stream archives count toward the 3,000-hour (Stage 1) and 8,000-hour (Stage 2) requirements. On this channel that means two kinds of content: regular finished long-form videos (about four minutes) and archives of live broadcasts.

Week	Regular long-form: views / min / subs	Live: views / min / subs
Oct 5	155 / 69 / +0	19 / 20 / +0
Oct 19	109 / 60 / +0	227 / 545 / +1
Oct 26	354 / 185 / +2	1,490 / 2,769 / +8
Nov 2	271 / 136 / +4	1,652 / 2,837 / +7

Discovery paths (last 90 days)

Regular videos: subscribers’ home feeds 2,293 views for 1,091 minutes (0.5 min per view); search 1,190 views for 792 minutes (0.7). Live: home feeds 1.0 minute per view, search 2.5, related videos 4.4. Unlisted videos (41) added only 97 views and 17 minutes and do not count.

Two readings

Constructive: even from search, regular videos earn 0.7 minutes, so topic and length may close part of the gap to live. Cautious: minutes per view fell from 0.97 (week of May 11) to 0.50, and 49 same-format videos in one week may have made individual titles harder to choose.

Next

The Nov 14-15 roundups test whether consolidated programs earn more minutes per view. Only six standalone long-form videos are planned next week.

YouTube ライブ ① 題材の格と、人が集まる場面

Why topic tier matters more than channel size

NFC Market Live is an AI-run channel that explains Japanese and US economic releases and US corporate earnings. Its monetization goal depends on public long-form watch hours, and public live archives count toward that total, so each live stream doubles as a long-form asset. The first YouTube monetization tier asks for 500 subscribers and 3,000 watch hours over 12 months; the channel stands at 198 subscribers and 217 hours.

The tier effect

• Apple/Palantir (large-cap): 2,076 minutes in seven days, 988 views, 5 new subscribers.
• AMD/Super Micro and Qualcomm/Arm (mid-cap): 280 and 310 minutes, still partial at days 6 and 5.
• Cloudflare, US jobs report, Monday preview: 91, 51 and 27 minutes.

Subscriber count at broadcast time barely differed (176 to 193), which points to subject matter rather than audience size as the driver.

Concurrent viewers

Large-cap viewership climbed after the filing and peaked at 30 during the cross-company analysis. The AMD/Super Micro stream peaked at 8 while waiting for the filing, then fell. The backstage review blames a slow first report: ‘cut the time from detection to first report to under one minute’.

How to read it

Interest in mega-cap earnings tends to concentrate on a handful of names, but this data alone cannot prove that. It is one large-cap broadcast, and day of week and timing cannot be separated from topic. NVDA on Nov 18 is the next test: does it exceed 1,000 minutes in seven days? Compare also whether CRWV and CSCO/AMAT land near the 280-310 minute mid-cap range.

YouTube ライブ ② 見つけられ方と、費用・品質

Discovery, cost and broadcast quality

How viewers find live streams

Over the last 90 days, live streams drew 2,135 views from subscribers/home (1.0 minute per view), 1,185 from search (2.5 minutes), 101 from related videos (4.4) and 66 from external sites (4.8). Search brings fewer but more purposeful viewers: for Apple/Palantir, search delivered 1,104 minutes against 728 from subscribers. For the US jobs stream, six external views produced 21 minutes, possibly traffic from X, though the sample is tiny.

Cost is flat, outcomes are not

Production cost per live stayed between roughly 1,000 and 1,600 yen, while seven-day viewing ranged from 27 to 2,076 minutes. Overall, live costs 202 yen per viewing hour, with earnings at 132 yen against 2,143 for indicator streams and 2,312 for the Monday preview. For context, US peers such as financial news channels spread fixed studio costs over far larger audiences; this channel’s cost is mostly LLM calls and voice synthesis, so it scales differently.

Quality issues

The Cloudflare stream had a 580-second silence; Apple/Palantir had four silences totalling 510 seconds. CPU peaked at 92-98% on four broadcasts, and the jobs stream had 6 voice dropouts out of 126 checks. The backstage review says: ‘stop other production jobs during broadcasts’ and ‘fill long silences with filler segments sooner’.

What to watch

The US CPI live (Nov 10), CRWV (Nov 11) and CSCO/AMAT (Nov 13) will show whether silences shorten once parallel jobs stop. A caveat: large-cap broadcasts have more concurrent viewers, so the same silence costs more there.

X ① 入口になった投稿の形と、クリックの出どころ

What opened the door on X

Scale this week

The channel posted 172 times on X this week, drawing 23,393 impressions, 187 link clicks, 41 profile clicks and 62 likes. The prior week had 29 posts, 1,354 impressions and zero clicks. Because the volume and formats changed so much (live-blogging during earnings and the jobs report began), a simple week-on-week comparison is misleading.

Post types

Impressions per post: first report 611, coming-up 518, stream announcement 422, detail index 217, detail 165, day-before notice 137, summary with four images 96 (Japanese) and 37 (English), shorts 86 (Japanese) and 52 (English).

The click pattern

Only ‘coming-up’ posts produced clicks at scale: 127 clicks across six posts (38 for AMD, 56 for Arm, 10 SMCI, 12 NET, 11 jobs, 0 Qualcomm). First-report posts carry links too, but drew zero clicks despite larger impressions, including 2,704 for the jobs report.

Context for non-Japanese readers

This resembles how US news accounts post a ‘coming up’ alert before a release and then a headline card: the alert drives traffic to the live stream, the headline card is consumed in-feed. Click counts exist only for posts with links, so zero on a link-free post means nothing.

Caveats

Six posts per group is a small sample, and impressions are the 24-hour value from the X API. The jobs first-report spike may reflect the jobs report’s wide audience rather than post design. Next test: CPI live on Nov 10 and CRWV on Nov 11.

X ② ライブへの呼び込みと、フォローへの結び付き

X live audience and follower conversion

Background for non-Japanese readers

The channel streams simultaneously on YouTube and X. X does not expose live viewer counts through its API, so the station director copies unique viewers and watch minutes by hand from X Media Studio. The X rewards program (Original Content Rewards) requires Premium, 500 verified followers and 500,000 impressions among verified users over 90 days. The channel is Premium but has 17 verified followers out of 281, and verified-user impressions cannot be pulled by API.

Live audience

• Nov 3, Apple/Palantir: 6 unique viewers, no live-blog posts.
• Nov 4, AMD/Super Micro: 517 unique viewers, the first day with live-blog posts.
• Nov 5, Arm/Qualcomm: 269. Nov 6, Cloudflare: 241. Nov 6, jobs report: 264.
• Oct 21: about 180, a rough figure.

Viewing per person was 0.7 to 2.0 minutes after live-blogging began. The jobs report ends almost instantly, which caps viewing time.

Followers

Followers went 275, 275, 282, 281, 281 from Nov 5 to Nov 9. Profile clicks across all posts this week: 41, led by coming-up posts (12), stream announcements (7) and first reports (5). Reaching 500 verified followers at the current 6.0% verified ratio would need about 8,265 followers overall.

Interpretation

The jump from 6 to 517 viewers suggests live-blogging works as an entry point, but one week cannot separate it from the topic mix. Conversion to follows is thin. Note that automated posting may not even qualify for the rewards program, which is why this is shown only as distance.

Next

Track the Nov 10 CPI live and Nov 11 CRWV broadcast: do unique viewers stay above 200, and does per-person viewing extend?

投資対効果と整理の候補

Return on spend: what the numbers say

Actual spend: 24,212 yen

Programs 9,230 yen, live 9,442 yen, shared processing 2,187 yen and trials/development 3,353 yen. Trials are excluded from the cost-per-viewing-hour calculation, leaving 20,859 yen and 335 yen per hour. Live costs 202 yen per viewing hour versus 741 yen for programs, whose shorts earned 13.1 viewing hours but only 2.3 hours came from long-form.

Why shorts do not help the first goal

YouTube’s first monetization tier counts only public long-form viewing and public live archives. Shorts do not count toward the 3,000 hours, which is why finished earnings programs look efficient on subscribers (287 yen each) but little on the hours goal.

Three-week trend

This week 24,212 yen bought 61.4 hours and 17 subscribers (395 yen per hour). Last week: 26,508 yen, 84.2 hours, 29 subscribers (315 yen). Two weeks ago: 19,531 yen, 14.7 hours, 1 subscriber (1,332 yen). Efficiency tracks how many mega-cap earnings land in the week.

Consolidation candidates

• Review US trade deep dive cost: 1,547 yen for 0.6 hours versus 200-400 yen for other deep dives.
• Drop the standalone long-form attached to the Monday live: 1,040 yen, 0.5 hours, zero subscribers.
• Limit shorts for Japanese and minor indicators to major release days.

Do not cut on numbers alone

Indicator lives (2,143 yen per hour, but the strongest X reach at 5,694 impressions), other-country deep dives (coverage), the BCB Focus long-form and trial earnings names. Like a wire service, breadth is part of the product, and YouTube can flag mass-produced repetitive content when reviewing monetization, so the director weighs both. The decision rests with the director.

来週の資本配分（まとめ）

Next week’s allocation in context

What is on the calendar

US CPI live on Nov 10 (22:15 JST, release at 22:30), CRWV live on Nov 11, a combined CSCO/AMAT live on Nov 13, and the Monday preview on Nov 16 built around Nvidia’s earnings on Nov 18. Four shorts are planned: CPI, CRWV, CSCO and the Nov 12 Japan securities-flow short. The weekly projection from the programming analysis: 8-18 long-form hours and +8-15 subscribers in the week of Nov 9, 20-40 hours and +15-30 for Nov 16 (Nvidia), and 2-4 hours and +3-6 for Nov 23, the first quiet week with only a Monday live.

Why these shifts

Big-cap earnings lives are the engine: one Apple/Palantir broadcast produced 34.6 of the week’s 49.5 long-form hours. Mid-cap lives bring 280-310 minutes. Pre-release posts with links produced 21.2 clicks each, so they stay. On the reduce side, the Monday live’s standalone long-form (1,040 yen, 0.5 hours) and costly deep dives are candidates. Programs serving distinctiveness or coverage, such as the indicator live and the BCB Focus long-form, are not cut on numbers.

The director’s calls

The director adopted the Monday live rebuild, the CPI live wrap-up that sets Japanese and US inflation documents side by side, and the weekend US macro merge. He declined extra shorts for AMAT, a Nebius live and moving BCB Focus into a roundup. The pattern: consolidation proposals pass more easily than adding programs for second-tier subjects.

What we will measure

Whether Nvidia exceeds 1,000 minutes in seven days, the quiet-week baseline of Nov 23, and whether silences shrink after parallel jobs stop during broadcasts. Viewer feedback is welcome in the comments.