
Mission 6 of 6
Grow your channel on four numbers and make it pay.
Stop refreshing analytics and read them properly in twenty minutes a week. Then learn how each income line works, with honest maths and a cadence that survives a day job.
What you'll do
Learn what CTR, average view duration, average percentage viewed and returning viewers each point at, and which mission fixes it.
Make one decision a week, post at a pace you can sustain with a job, and turn long videos into Shorts.
AdSense, sponsorships priced from your own stats, affiliates and your own product, in the order they usually arrive.
- A script that diagnoses every video
- A rate card built from your own numbers
- Rules for sponsors and affiliates that protect your viewers' trust
- At least three published videos with a week of data each. Fewer and the numbers are noise; do the review anyway for the habit
- YouTube Studio access with Advanced mode analytics
- Python 3
- An idea of what you'd sell or send people to. If it's nothing yet, the last step points at two missions that build it
Your progress0/6
- 01The four numbers and what each one points at12 min
- 02The weekly review: twenty minutes, one decision10 min
- 03Cadence: weekly, batched, survivable8 min
- 04One long video into a week of Shorts that point back8 min
- 05The money: AdSense, sponsors from your own stats, affiliates, your own product16 min
- 06The next ninety days, and what to build to sell6 min
The four numbers and what each one points at
Every other metric in Studio is a consequence of these four. Learn to read them as diagnoses, not scores.
Four numbers matter in Studio, and each one points at a specific fix in an earlier mission.
Open Studio, Analytics, Content, and look at one video.
Ignore views for now. Views are the output; these four are the causes.
Impressions click-through rate (CTR)
Of the people shown the thumbnail, how many clicked. This is packaging and nothing else: title, thumbnail, and whether the two match the viewer the video was shown to.
YouTube's own help says half of all channels sit between 2 and 10 percent, and that CTR falls as a video gets shown to wider, colder audiences, so a dropping CTR on a video that's growing can be fine.
A low CTR from day one is a thumbnail problem. Fix in the packaging mission; the thumbnail test gives you the evidence.
Average view duration (AVD)
How long the typical viewer watched, in minutes. The number YouTube weighs most for suggested traffic, because it's session time.
Across your channel it should go up as the scripts get tighter. A high AVD on a long video is the goal; two 6-minute videos with 3 minutes AVD each are not the same as one 12-minute video with 6.
Average percentage viewed (APV)
AVD divided by length. This is the script and the edit.
It tells you whether the problem is the cold open (a cliff at the start), the middle (a steady slide, missing payoffs) or the end (a drop before the next-video handoff). Open the retention graph next to it; the shape says which.
Fix in the script mission, or the edit if the cuts are slow.
Returning viewers
You'll find this in the Audience tab: of the people who watched this week, how many had watched before. This is the channel promise.
If it's rising, the videos are about the same thing and people know what to expect. If it's flat while views grow, you're getting one-off search traffic from videos that don't connect, which is fine money and poor growth.
Fix in the first mission: the promise and the plan.
What to ignore for now
Two numbers people stare at that you can mostly ignore for now: subscribers (a lagging effect of returning viewers) and impressions (YouTube's decision, driven by the four above).
And one to add when it exists: the 30-second retention number from the packaging mission, which is the fastest single check on a new upload.
Let a script do the diagnosis
Export the Video table from Advanced mode as a CSV, with CTR, APV and AVD columns added.
Save this script, which reads the export and ranks your videos on the four numbers.
review.py# review.py: read a YouTube Studio "Table data" export and rank videos on the four numbers # Studio > Analytics > Advanced mode > Video tab > Export (CSV). usage: python review.py "Table data.csv" import csv, sys def col(headers, *needles): for h in headers: if all(n.lower() in h.lower() for n in needles): return h sys.exit(f"column not found: {needles}. Add 'Impressions click-through rate' and 'Average percentage viewed' in Studio before exporting.") rows = list(csv.DictReader(open(sys.argv[1], encoding="utf-8-sig"))) h = rows[0].keys() title, views = col(h, "title"), col(h, "views") ctr, apv = col(h, "click-through"), col(h, "percentage viewed") avd = col(h, "average view duration") rows = [r for r in rows if r[title] and r[title].lower() != "total"] def num(x): try: return float(str(x).replace("%", "").replace(",", "")) except ValueError: return 0.0 print(f"{'video':<48}{'views':>8}{'CTR%':>7}{'APV%':>7}{'AVD':>9} diagnosis") for r in sorted(rows, key=lambda r: -num(r[views])): c, a = num(r[ctr]), num(r[apv]) diag = "packaging" if c < 3 else ("script/edit" if a < 35 else "working: make more like this") print(f"{r[title][:47]:<48}{num(r[views]):>8.0f}{c:>7.1f}{a:>7.1f}{r[avd]:>9} {diag}")Run it on your export.
Terminalpython review.py "Table data.csv"You get one line per video with its views, CTR, APV and AVD, and a diagnosis at the end: packaging, script/edit, or working: make more like this.
The thresholds in it (3 percent CTR, 35 percent APV) are starting points for a new builder channel, not laws.
After ten videos, replace them with your own medians: a video below your median CTR has a packaging problem relative to your own audience, which is the comparison that matters.
WarningDon't read any of this on a video under 48 hours old or under a few hundred views. Early numbers come from subscribers and your Instagram crossover, who behave differently from the cold viewers YouTube tests on later.