Rogue Canon · Weekly analysis · Studio export + Analytics API

Week in Review · 4–11 July

19,448 views · 149 watch-hours · +145 subscribers (158 total) · and the promo question, answered with numbers.

The verdict you asked for: stop the promos. All of them.

This week's data closes the case. Paid views don't hold (promoted Shorts averaged 3–8 seconds per view), paid subs don't watch (of 145 new subscribers, 144 came from the ad unit and only 1 from actually watching a video), and your organic engine just showed its first real pulse without any help. There is exactly one defensible future spend — it's at the bottom, and it can wait until 21 August.

What the money actually bought

SignalNumberRead
Ad views (7d)5,927 raw → 2,616 engaged56% of paid plays are instant skips
Promoted Shorts hold3.4–8.3s average viewPaid swipe traffic — worthless for the algorithm
Long-forms under paid mix~34% held vs 53–59% organic lifetimePaid viewers dilute your retention stats
Subscribers+145; 144 ad-unit, 1 content-earnedReal accounts, low intent — subscriber-source views were just 25 all week

The strategic cost nobody invoices: YouTube learns from your audience. A subscriber base that never watches teaches the algorithm your content doesn't convert — which suppresses exactly the organic reach we're trying to build.

The organic pulse — it's started

Source (7d, engaged views)ViewsWatch min
Shorts feed — first real showing5416
Playlists (deepest: 2.5 min/view)41104
Subscribers2544
Suggested + search + other~32~47

~152 engaged organic views this week — versus ~128 in the entire previous 28 days. The Shorts feed opened its door (a row that didn't exist at all last month). This happened before the hook-first cuts, the chant sheets, or the matchday engine exist. The levers we've built press directly on the things now moving.

The overlooked win: your CTR data

VideoImpressionsCTRAction
Pressure Makes Warriors678.96%Thumbnail works — needs surface area
Loose Cannons Live2238.52%Same — excellent packaging
Built Different1906.84%Healthy
Midnight Run330 (most shown)3.03%Run your prepared A/B thumbnail test
Welcome To The Crest2861.75%Weakest — test it too

Anything above ~4–5% CTR at these impression levels is genuinely good. Your two most-shown videos convert worst — and you already made the A/B thumbnails (they're sitting in the roguecanon folder with a documented test plan). This is the highest-leverage free move available: YouTube Studio → each video → Test & Compare → upload the three prepared thumbs.

Next 7 days, ranked by leverage

1

Pause all four active promos

Today. Every day they run adds swipe-junk to your stats and teaches the algorithm the wrong lesson.

2

Launch the thumbnail tests (Midnight Run + Crest)

Studio → Test & Compare with your prepared A/B/C sets. 600+ impressions/week are currently converting at 2–3%.

3

Deploy roguecanon.com

Chant sheets + nav are staged. This is the site-engagement engine — then I push chant-sheet links into descriptions and pinned comments by API.

4

Cut and post S1 (We Don't Break matchday cut)

The Shorts feed just proved it will carry you — 54 engaged views on old cuts. Give it the hook-first format.

5

Community poll Sunday

158 subscribers now — most brand new. One poll converts passive ad-subs into invested voters.

The one promo I'd ever bring back

The case: a single £50–75 subs-goal burst (Underdog/Loose Cannons creative, ~29p/sub) in the week before the season opener (Fri 21 Aug) — if and only if crossing 1,000 subscribers before kickoff matters for social proof and unlocking YPP review sooner. That's a branding argument, not a growth argument: paid subs won't watch, and the 4,000 watch-hours YPP requires can only come from real viewers anyway.

My recommendation: hold even that. Decide in mid-August with six weeks of organic data. If the matchday engine is compounding, you won't want it; if it isn't, we'll know exactly what to fix instead of papering over it with spend.