Esekai, vertical video production agency Book a call
←︎ All articles Reels & social 12 min · 21 November 2026

YouTube Analytics: The Metrics That Matter for a Brand

Click-through rate, retention, “Stayed to watch”, traffic sources, returning viewers: how a brand should read YouTube Analytics for Shorts and long videos, and what to do next.

Tony Lo, founder of Esekai
By Founder and CEO of Esekai, Paris

For a brand, YouTube Analytics should be read through six metrics, not the view counter: thumbnail impressions and click-through rate, average view duration, audience retention (key moments), the share of viewers who “stayed to watch” a Short, traffic sources, and the balance between new viewers, returning viewers and subscribers gained. The view counter, for its part, has become misleading: since 31 March 2025, a Short view is counted as soon as playback starts or restarts, with no minimum duration. Everything that follows is based on YouTube’s official Help Centre, consulted in November 2026.

What Is YouTube Analytics, and Where Are the Useful Figures?

YouTube Analytics is the dashboard built into YouTube Studio: it measures the reach, engagement and audience of every channel and every piece of content. The difficulty is not accessing the figures, but knowing which ones to look at among the dozens on display.

Since the Content tab was redesigned, YouTube separates reports by format: videos, Shorts and live streams do not have the same key metrics card. For videos, the card highlights impressions, click-through rate and retention; for Shorts, views, likes and subscribers. This separation is already a lesson in itself: a Short and a long video are not judged by the same figures.

Why Is the View Count No Longer Enough to Judge a Video?

Because the definition of a view has changed, and it says nothing about what people actually watched. On 31 March 2025, YouTube changed how Shorts views are counted: according to TechCrunch, a view is now counted every time a Short starts or restarts playing, whereas previously it had to be watched for a certain number of seconds. The old metric has not disappeared: it is now called “engaged views” and can be found in advanced mode. It is also the one used to calculate revenue and eligibility for the YouTube Partner Programme.

YouTube Help now confirms this for all formats: views are counted as soon as playback begins, Shorts included. And since November 2026, public counters include co-viewing, meaning an estimate of people watching together on a TV screen. Engaged views, on the other hand, exclude loops and co-viewing: they are the only basis for comparing a Short from 2024 with a Short from 2026.

There is a deeper reason not to stop at views. In a post on its official blog, YouTube’s VP of Engineering explains how recommendations work: the platform learnt as early as 2011 that a click does not mean someone watched, and it built watch time into recommendations in 2012. Views immediately dropped by 20%. A video that attracts clicks without holding attention is therefore pushed less, which makes watch metrics more telling than the counter.

Which Metrics Should You Track, Format by Format?

For long videos, the trio of impressions, click-through rate and retention; for Shorts, the share of viewers who stay to watch and engaged views; for the channel, traffic sources and audience loyalty. The table sets out the official definitions and the question each metric helps you answer.

Metric (Studio label)Official definitionFormatThe question it answers
Thumbnail impressionsNumber of times your thumbnails were shown on YouTube, through logged impressionsLong videos, live streamsIs YouTube showing my video, and to how many people?
Impressions click-through rateHow often viewers watched a video after seeing its thumbnailLong videos, live streamsDo the title and thumbnail make people want to click?
Average view durationAverage minutes watched by those who stayed, calculated from engaged viewsAllHow long does the video really hold attention?
Key moments for audience retentionCurve and markers (intro, top moments, spikes, dips) of attention throughout the videoLong videosAt exactly what point do people leave?
Stayed to watchPercentage of times viewers watched beyond the first few seconds of a ShortShortsDoes the hook stop the thumb?
Engaged viewsNumber of times viewers watched beyond the first few seconds, excluding loopsMainly ShortsHow many genuine plays, measured by a consistent rule?
Traffic sourcesHow viewers found the content, on YouTube and elsewhereAllWhat brings me an audience: search, suggestions, the Shorts feed?
New viewers, returning viewers, subscribersFirst visit, repeat monthly visits, subscriptions gainedChannelAm I building an audience or renting attention?

How Do You Read Thumbnail Click-Through Rate Correctly?

By always reading it alongside the number of impressions and the traffic source, never on its own. According to the official FAQ on impressions and click-through rate, half of all channels and videos on YouTube have an impressions click-through rate of between 2% and 10%. Videos and channels less than a week old, or with fewer than 100 views, may fall outside this range. Not all impressions are counted: those from external websites and end screens are excluded, and seeing a thumbnail does not necessarily generate a logged impression. That is why dividing views by impressions does not give you the click-through rate.

A Falling Click-Through Rate Can Be Good News

YouTube gives a worked example on its page about interpreting click-through rate: a video gets 10,000 impressions with a 9% click-through rate, mainly among loyal fans. A week later, it has reached 100,000 impressions and a 3.5% click-through rate. The rate has fallen, but the video has been shown to 90,000 more people: it is reaching beyond its core audience. The same page points out that search generally brings fewer impressions but a better click-through rate, because intent is specific, whereas the home page brings many impressions with a lower rate.

The Warning Sign: Clickbait

YouTube’s FAQ describes the signature of a misleading thumbnail: a high click-through rate, a low average view duration and fewer impressions than expected. These videos are less likely to be recommended.

What to Do

  • Wait for a significant volume of impressions before drawing conclusions, as YouTube recommends, and do not react to small fluctuations.
  • Compare videos over time and from the same traffic source: a video pushed on the home page and a video found through search do not have the same normal rate.
  • Test rather than guess. The title and thumbnail A/B testing tool lets you compare up to three titles, three thumbnails or three combinations. The winner is the one that achieves the most watch time, not the best click-through rate. The test runs for up to two weeks, is launched from a computer, and is not available for Shorts, scheduled live streams or Premieres.

How Do You Read Audience Retention for a Long Video?

The retention curve shows, second by second, what share of viewers is still watching. You read it by looking for breaks, not the average. YouTube’s help page on key moments for audience retention provides the reading guide: a flat line means people are watching that section from start to finish, a gentle slope that they are gradually losing interest, which is the norm since videos generally lose viewers as they play. A spike indicates a section that was watched, rewatched or shared; a dip, a section that was skipped or abandoned.

The report breaks the video down into four types of moment:

MomentWhat it measuresWhat to do about it
IntroThe share of the audience still there after the first 30 secondsIf it is low, check that the opening delivers on the promise of the title and thumbnail, then test a different opening
Top momentsThe sections where almost nobody drops offIf they come late, move this type of content earlier in the video
SpikesSections that were rewatched or sharedEither the section is valuable and deserves a format of its own, or it was confusing and had to be rewatched
DipsSections that were skipped or where people stop watchingCut, shorten or rewrite these sequences in future videos

Moments are only detected if the video is at least 60 seconds long and has more than 100 views. And the “typical retention” shown for comparison corresponds to your last 10 videos of similar length: it is your own benchmark, not a YouTube average.

Shorts: Which Figures Should You Look at, Given There’s No Click-Through Rate?

On Shorts, there is no thumbnail to click in the feed: the first filter is the “Stayed to watch” metric, followed by engaged views, average view duration and subscribers gained. YouTube defines this metric as the percentage of times viewers watched beyond the first few seconds of a Short. It plays the role that click-through rate plays for a long video: it tells you whether the hook is working.

A useful reading takes place in three stages:

  • Low “Stayed to watch”: the problem lies in the first frame and the first few words. Rework the hook, not the rest of the video.
  • Good hook, but low average view duration: the video is losing people along the way. For Shorts, average duration is calculated from engaged views, which makes it an honest measure of the middle of the video.
  • Good retention, few subscribers gained: people like the content, but nothing makes them want to follow the channel. A recognisable series or a regular slot often makes all the difference.

Reach reports also list traffic sources specific to Shorts: the Shorts feed, sound pages and remixed videos. This is precisely the work of a production designed for this format, which we explain in detail on our YouTube Shorts agency page.

Where Do Your Viewers Come From, and What Can You Learn?

The traffic sources report tells you which YouTube mechanism is working for you: search, suggestions, the home page, the Shorts feed or external websites. Each source calls for a different action. The definitions below come from the help page on reach.

SourceWhat it coversWhat a brand can take from it
Browse featuresHome, Subscriptions, Watch Later, Trending and other browsing areasIf it dominates, the algorithm is recommending the video: polish the thumbnail and retention to extend its reach
Suggested videosSuggestions alongside or after other videos, and links in descriptionsIdentify the videos sending you traffic and produce related topics
YouTube searchSearch results, with the search terms reportUse the actual search terms in the titles and descriptions of future videos
YouTube advertisingVideo used as an adIsolate this traffic before judging organic performance
External, direct or unknown sourcesWebsites and apps that embed or link to the video; direct entry, bookmarks, logged-out viewersMeasure the impact of your newsletters, your website and your other social networks

Advertising often skews the figures. According to the same page, a skippable ad counts as a view when it lasts more than 10 seconds and is watched for 30 seconds or to the end; a non-skippable ad never counts. If your brand buys advertising on one of the channel's videos, filter out that source before comparing it with your organic videos. Creating the ads follows a different logic, which we cover in detail on our YouTube Ads agency page.

New Viewers, Regular Viewers, Subscribers: How Do You Measure a Returning Audience?

YouTube groups viewers by how often they visit, which lets you tell a video that brings in new people from one that feeds your loyal audience. The definitions in the Audience tab help page are precise:

  • New viewers: people watching a video from your channel for the first time in the period. Someone who hasn't visited for more than a year counts as new.
  • Casual viewers: at least one visit a month, for one to five months over the past year.
  • Regular viewers: at least one visit a month, for more than six months over the past year.
  • Subscribers: the number of viewers who have subscribed to the channel.

A channel that lives solely on new viewers depends on every single upload; a channel that only speaks to its regulars hits a ceiling. The trend matters more than the figure: if the share of regular viewers grows quarter after quarter, the channel is building an asset.

The report on the formats your audience watches, calculated over the last 28 days, settles a common question: should you invest in Shorts? If your audience already watches plenty of them, the answer leans towards yes.

What Should You Actually Do With These Figures, Week After Week?

Turn every reading into a production decision: change a hook, a thumbnail, a length or a topic. Here is a simple routine to get there.

  1. Two days after each upload: check "Stayed to watch" for Shorts, and the intro at 30 seconds for long-form videos. This is your hook diagnosis.
  2. After one to two weeks: read the click-through rate alongside impressions and traffic source, and check the result of the A/B test if you ran one.
  3. Every month: note the retention dips shared by several videos, the search terms driving traffic and how your regular viewers are evolving.
  4. Every quarter: compare formats against each other on engaged views and watch time, then reallocate production effort towards what builds your audience.

Record every decision and its outcome: after a few months, this log is worth more than any industry average. To fit it into a full editorial calendar, our 6-month social media plan provides the framework.

Frequently Asked Questions

What Is a Good Click-Through Rate on YouTube?
YouTube states that half of all channels and videos have an impressions click-through rate between 2% and 10%. The rate varies depending on the content, the audience and where the thumbnail is shown.
Why Do My Shorts Get More Views Since 2025?
Since 31 March 2025, a Shorts view is counted as soon as playback starts or restarts, with no minimum duration. The old count still exists under the name "engaged views", in YouTube Analytics' advanced mode. That is the one to use when comparing Shorts published before and after the change.
What Is the "Stayed to Watch" Metric?
It is the percentage of times viewers watched a Short beyond its first few seconds instead of swiping to the next one. It measures the strength of the hook.
How Can I See When Viewers Leave My Video?
Open the video in YouTube Studio, then the key moments for audience retention report. It shows the intro (the share of viewers still watching after 30 seconds), top moments, spikes and dips. The video must be at least 60 seconds long with more than 100 views, and the data arrives within one to two days.
Can You Test Several Thumbnails on YouTube?
Yes, with YouTube Studio's A/B test: up to three titles, three thumbnails or three combinations. YouTube picks the option that generates the most watch time, and the test usually ends within two weeks at most. It is not available for Shorts, scheduled live streams or Premieres.

Do Your YouTube Figures Point to the Hook?

Retention that collapses in the first few seconds, or Shorts that people swipe past, can't be fixed in a dashboard: they are fixed during the shoot and the edit. Esekai is a vertical video studio based in Paris 8e, with 20 people, a casting pool of over 300 creators and two hooks shot for every video, so you can test which one holds attention. More than 350 brands have trusted us with their videos. Thirty minutes on the phone: formats, budget, action plan.

Book a call →

30 minutes to discuss formats, budget and an action plan.

Read next

Reels & social16 min YouTube Shorts for a brand: is it worth it? The real maths Creative ads1 min A Social Media Editorial Line: The Secret Weapon of Brands That Last Creative ads17 min iPhone or camera shoots for social: the threshold that decides TikTok17 min TikTok Content Agency for Brands: The 5 Trades Behind the Term