The numbers everyone assumes you know.

CTR is impressions divided by clicks, shown in YouTube Studio as a percentage. If 1,000 people saw your thumbnail in their feed and 40 clicked, that is a 4% CTR. It measures your packaging — the thumbnail and title — not the video itself.

Why it matters: Most established channels sit between 2% and 10%. Below ~2% the packaging is usually the bottleneck, not the content; above ~10% you are often being shown to a very warm audience (subscribers, or a small niche). CTR and retention compound: a high CTR on a video people abandon will suppress future impressions.

YouTube reports this two ways: average view duration in minutes and seconds, and average percentage viewed. The retention graph shows it moment by moment, so you can see exactly where people leave. A cliff in the first 30 seconds is a hook problem; a slow steady decline is a pacing problem.

Why it matters: Retention is the signal that decides whether YouTube keeps showing a video after its first push. The first 30 seconds carry disproportionate weight — that is where most channels lose half their audience — which is why editing, not filming, is usually where retention is won.

RPM is your total revenue divided by total views, times 1,000. It includes everything YouTube pays you — ads, memberships, Super Chat — and it is calculated after YouTube takes its share. It is the number that matters to you, unlike CPM, which is what advertisers pay before the split.

Why it matters: RPM varies enormously by audience geography and topic — the same video can earn several times more from a US audience than an Indian one. That is why views and income often move independently, and why a channel can grow while earnings stay flat.

CPM is the advertiser side of the transaction. Creators frequently quote CPM when they mean RPM, which makes earnings sound higher than they are. Your RPM will always be meaningfully lower than your CPM, because YouTube keeps a share and not every view carries an ad.

Why it matters: If someone tells you their channel's earnings using CPM, the real take-home is materially lower. Compare channels on RPM or nothing.

YPP is what you join to run ads and use most monetisation features. YouTube's published bar for ad revenue is 1,000 subscribers plus either 4,000 valid public watch hours in the past 12 months or 10 million valid public Shorts views in 90 days, along with following the platform's policies.

Why it matters: The subscriber count is rarely what blocks people — watch hours are. That makes long-form content the faster route into YPP for most channels, since Shorts views convert to watch hours very inefficiently.

Demonetization comes in two very different forms. A single video can be limited or have ads disabled for advertiser-friendliness reasons, which is common and often fixable. Whole-channel removal from YPP is far more serious and usually follows repeated policy strikes, reused content findings, or inauthentic activity.

Why it matters: The two cases need opposite responses. A per-video limit is usually a metadata, thumbnail or content-edit problem you can appeal. Channel-level removal needs a genuine content and process change before reapplying — reapplying unchanged almost always fails.

An impression is counted when your thumbnail is displayed on-screen long enough to register — in the home feed, in suggested videos, in search. It is not the same as a view, and it is the denominator underneath your CTR.

Why it matters: Falling views with steady impressions means a packaging problem. Falling impressions means YouTube itself has pulled back distribution, which is usually a retention or freshness signal rather than a thumbnail issue. Checking which one moved tells you what to fix.

Watch hours aggregate every second watched across your public videos. YPP counts valid public watch hours from the trailing 12 months, so old watch time expires out of the window as it ages.

Why it matters: Because the window rolls, a channel that stops publishing can drop back below the threshold. Consistency matters for eligibility, not just for growth.

The hook is the first few seconds of a video — usually the first 5 to 15 — where you establish what the viewer is about to get and why it is worth their time. In editing terms it is often built by cutting away the greeting entirely and opening on the most interesting moment.

Why it matters: This is the single highest-leverage edit on most videos, because the largest retention drop of the whole video happens here. Rewriting the first 15 seconds routinely moves average view duration more than anything done later in the timeline.

Repurposing takes an existing long-form video and cuts self-contained vertical clips from it for Shorts, Reels and TikTok. Done properly each clip is re-hooked and re-captioned for a feed where there is no context and no patience, rather than simply cropped.

Why it matters: It is the cheapest way to raise output, because the filming is already paid for. The failure mode is lazy cropping: a clip that assumes the viewer saw the long version will underperform every time.

YouTube can rotate multiple thumbnails on the same video and report which performed best. The discipline is to change one variable at a time — face versus no face, three words versus six — so the result actually tells you something.

Why it matters: Opinions about thumbnails are close to worthless; measured clicks are not. Testing one variable per upload builds a set of rules specific to your audience rather than generic advice.

On Shorts, Reels and TikTok, captions, usernames, buttons and progress bars sit on top of your video. Safe zones are the regions — roughly the middle of the frame, away from the bottom third and right edge — where your own text and subject stay visible.

Why it matters: Text placed outside the safe zone gets covered on some devices and not others, so it can look fine in your editor and be unreadable for a large share of your audience.

Content ID is YouTube's automated system for matching uploaded audio and video against a database of registered material. When it matches, the rights holder chooses what happens: monetise the video and take the revenue, block it in some countries, or just track its stats. You can dispute a claim if you have the rights or a fair-use case.

Why it matters: A Content ID claim does NOT put your channel at risk — it affects one video's revenue or availability, nothing more. Creators routinely confuse claims with strikes and panic, or delete videos they did not need to. The two are completely different systems with completely different consequences.

A strike is issued when a copyright owner submits a legal takedown request and YouTube removes the video. Unlike a Content ID claim it is a channel-level penalty, and it comes with restrictions such as losing access to live streaming for a period. A strike expires after 90 days, and completing Copyright School is required.

Why it matters: Three copyright strikes active at once means the channel is terminated and the account cannot create new channels. That is YouTube's published rule, and it is the single fastest way to lose a channel outright — which is why distinguishing a strike from a claim is not pedantry.

In YouTube Analytics, Browse features means your video appeared on someone's home feed or subscriptions; Suggested videos means it appeared beside or after another video. Both are YouTube choosing to show you to someone, rather than someone searching for you.

Why it matters: A channel growing mainly from Search has a ceiling set by how many people search that term. A channel growing from Browse and Suggested does not have that ceiling, which is why packaging work — thumbnail and title — pays off disproportionately: those surfaces are won on the click, not the query.

End screens are elements YouTube lets you place over the final 5 to 20 seconds of a video, linking to another video, a playlist, or a subscribe button. The video has to be at least 25 seconds long to carry one, and you need to leave visual room for them in the edit.

Why it matters: An end screen only works if the edit plans for it. If your video cuts to black the instant you stop talking, the cards sit on an empty frame and nobody clicks. Leaving 10 to 15 seconds of designed outro is an editing decision, not a post-production afterthought.

B-roll is any secondary footage — screen recordings, product shots, stock clips, cutaways — layered over your primary audio. It covers edits, illustrates what you are describing, and gives the eye something new without breaking the sentence.

Why it matters: B-roll is the main tool for fixing a slow-declining retention graph. Attention drops when the frame stops changing, so arriving with a cutaway at the moment interest would dip is usually more effective than cutting the segment out entirely.

YouTube Studio reports Shorts performance partly as viewed versus swiped away — how many people who were shown your Short stayed for it rather than scrolling on. It is the short-form equivalent of a thumbnail click plus a first-second hook, measured together.

Why it matters: Short-form is judged on completion, not length. A tight 20-second clip that people finish will out-perform a padded 58-second one they abandon at 15 seconds, which is why cutting a Short to fill time is actively counterproductive.

A stream package is the overlay set a streamer uses: in-stream overlay, starting-soon, be-right-back and ending screens, alert graphics, transitions, panels and an offline banner. It is built to the safe areas of OBS or Streamlabs and handed over as layered source files so it can be edited later.

Why it matters: A stream package is what makes a channel recognisable in a thumbnail-sized preview and across clipped highlights. Piecemeal overlays collected over time read as amateur precisely because nothing matches, which costs you the thing branding is for.

A lower third is the overlay that identifies who is speaking or labels what is on screen, placed in the lower area of the frame. In vertical formats it has to respect safe zones, because platform UI covers the bottom of the frame on Shorts, Reels and TikTok.

Why it matters: Lower thirds are where inconsistent branding shows up first — a different font or colour on every video is the most common reason a channel looks unprofessional despite good footage. One set, reused, fixes it.