The YouTube Analytics reporting dashboard for agencies

The subscribers tab carries a five-figure band — subscriber count, net new, gained, lost and growth rate — over a daily table that walks the subscriber level day by day with that day's gains, losses and net movement. Beneath it, subscribers gained are broken out by content format with a per-thousand-views rate, so a format that pulls large view counts and almost no subscribers is visible instead of assumed. YouTube credits a subscribe to the video that earned it, while most unsubscribes happen away from any video's page, so the report keeps losses at channel level rather than dividing them across videos to make a table look complete.

The second half of the tab is the split most channel reviews skip: how much of the watch time came from people already subscribed, and how much from people who were not. A channel taking almost all its watch time from non-subscribers is being carried by recommendation rather than by an audience, which is a fine month and a fragile position; one taking most of it from subscribers has a following that will show up for the next upload but may have stopped reaching anyone new. Both readings change what to publish next, and neither is visible in a subscriber count alone.

Every video that earned a view in the window gets a row: format, publish date, views, watch time, average view duration, average percentage viewed, impressions, click-through rate, and its likes and comments. Sort by views, by watch time or by newest, and open any row for the full card — subscribers gained from that video, end-screen and card clicks, views in its first seven days, and a bar showing how far into it the average view reached. Older uploads that are still being watched appear alongside the month's new ones, because a channel's month is rarely made only of what it published in it.

The retention tab then puts the two halves of a video's performance side by side. A distribution shows how many videos land in each retention band, per-video bars rank them, and a table sets impressions and click-through rate against average percentage viewed with a plain-language read of each row. A strong click-through rate with weak retention is a thumbnail promising something the video did not deliver — the most common and most expensive pattern on a channel, and one no single metric will ever surface. Rates are scored against the band the format itself sits in, never against a number borrowed from someone else's study.

Shorts, long-form uploads and live streams each get a card carrying the count in the report, views and share of views, watch hours and share of watch time, average view duration, average percentage viewed and subscribers gained, so each format is described on its own terms before any of them are ranked. A table underneath then compares them on averages per video — average views, average watch time, average view duration, retention, click-through rate and subscribers per thousand views — because totals simply reward whichever format happened to be published most often that month, and averages answer the question an account manager was actually asked in the review.

The distinction is not academic. A single Short can out-earn a quarter of long-form uploads on views while contributing a small fraction of the watch time and fewer subscribers than one mid-sized tutorial; a live replay collects a long runtime that almost nobody finishes. Average percentage viewed is what puts a forty-five-second vertical clip and a twenty-minute walkthrough on the same axis, since it measures how much of each video the average view actually got through. The content filter at the top re-bases the entire report — scorecards, trend, mix, and all eight tabs — so a Shorts-only or long-form-only reading is one click, not a second export.

Nine traffic sources are reported with views, share, watch time and — the column that usually decides the argument — the average view duration a view from that surface actually runs. YouTube search and playlists bring the longest views; the Shorts feed brings the shortest by a wide margin; suggested videos and browse features sit in between. The list orders by views or by watch time, and the two orderings frequently disagree, which is precisely the point: the surface delivering the most views is often not the one delivering the audience.

Because the mix is built from what the channel published rather than from a fixed template, the Shorts feed row rises and falls with the Shorts themselves and drops to zero when the report is filtered to long-form. Sitting under the search row, a table of the twelve YouTube search terms behind it shows each query's views and its share of search views. Those are queries typed into YouTube, which is a different behaviour from the queries a website ranks for in a search engine — that half of the picture lives in the search reporting alongside this page.

The audience tab reports the seven age bands YouTube publishes as stacked bars split by gender, new against returning viewers, and the top geographies, with a toggle that swaps every figure between views and watch time. Composition is reported for the channel and labelled as such, so nobody has to guess whether a demographic split has quietly been narrowed to one format. Toggling to watch time regularly reorders the list — an age band or a country can be a modest slice of the views and a large slice of the hours, which is the segment worth making the next video for.

The devices view is the one agencies forget. Watch time is broken out across mobile phones, computers, TVs, tablets and game consoles, and the TV row changes production decisions: on-screen text sized for a phone is unreadable in a living room, and a thumbnail competing in a ten-foot interface is a different design problem. Beside it, publishing cadence charts uploads and views by day of the week and reports the average views a new upload collected in its first seven days, so a publishing schedule can be argued from this channel's own record.