How-To Guide
How do you measure video completion rate on social media?
Fix three definitions before comparing platforms: a view is playback started, a completion reaches the end, and a rewatch is more than one play per viewer.
Divide the plays that reached the final second by the plays that started, then multiply by 100, and record that number beside the platform's own definition of a view, the video length and the average view duration. The arithmetic is easy; the comparison is the hard part, because YouTube, Instagram and TikTok start counting at different moments. Fix the denominator per platform, compare a video with its own past runs and with videos of the same length, and treat a rewatch as a separate check.
- A view is not a completion. [YouTube counts a view when playback starts](https://support.google.com/youtube/answer/2991785?hl=en) and [Meta counts a play as a View](https://creators.facebook.com/blog/introducing-views-to-simplify-content/?locale=en_US), so neither number says anyone reached the end.
- TikTok Ads publishes 25, 50, 75 and 100 percent buckets, so [its 100 percent video views divided by its video views](https://ads.tiktok.com/resources/help/article/video-play) is the cleanest native completion figure of the three platforms.
- Completion targets move with length. [Retensis advises above 40 percent under 15 seconds, above 30 percent at 15 to 30 seconds and above 25 percent at 30 to 60 seconds](https://retensis.com/blog/youtube-shorts-completion-rate-benchmarks-2026), as a vendor benchmark.
Why one completion number cannot compare platforms
A completion rate looks like one number, but it is built from two choices: what counts as a start and what counts as an end. Every major platform makes those choices differently. Hootsuite warns that every platform defines a view differently, so comparing raw view counts across TikTok, Instagram, YouTube and LinkedIn without context can mislead. The warning matters most for completion, because the completion share inherits whatever the platform called a start.
The start is a different event on each app. YouTube counts a public view the moment a video starts to play. Meta defines Views as the number of times a reel or video was played, and that includes repeat looks. TikTok Ads reports 2-second and 6-second video views as separate counts within an impression session. Dividing the same end count by those three starts produces three different rates.
The format changes the start too. Google Ads says the point where the video player is initiated differs by format, and an in-stream ad starts playing automatically. An autoplay placement can therefore log a start from a viewer who never chose to watch. Hootsuite's practical answer is to lean on retention and completion on autoplay feeds, where play rate loses meaning.
A blended cross-platform average hides all of this. Socialinsider notes that each platform has its own set of metrics and dashboards, which fragments the data. The workable rule is narrow: compute completion inside one platform, compare videos of the same length, and label any cross-platform table as a set of estimates with three different denominators.
Original diagram of the three numbers to fix before comparing platforms. Definitions follow YouTube Help, Meta for Creators and TikTok Ads Manager, read November 17, 2026.
Make it work for your brand
Find it. Remake it. Post it. Superdirector turns the viral formats in your niche into your brand's carousels, X posts and video scripts, and keeps the calendar full.
Define the three numbers first: view, completion and rewatch
Before any division, write three definitions on one line for each platform: what starts the count, what ends it, and how a repeat play is treated. This is the step that makes a later comparison honest, and it is the step most scorecards skip.
The view is playback started. That is the denominator. On YouTube it is the public view count, and YouTube keeps a separate engaged-view idea for Partner Program earnings, so never mix the two. On Meta it is Views, which count a reel or video play including repeat looks. On TikTok Ads it is video views, with 2-second and 6-second views as separate thresholds (TikTok Ads Manager).
The completion is playback that reached the end. TikTok Ads makes this the easiest of the three, with video views at 100 percent as its own metric (TikTok Ads Manager). YouTube shows average percentage viewed and a retention graph, and Hootsuite describes YouTube as framing completion as average percentage viewed instead of a binary completion. Meta does not publish a separate completion field, so an Instagram or Facebook completion is an estimate built from the watch-to-end data the dashboard exposes.
The rewatch is more than one play from the same viewer, and it is the number platforms have been quietly removing. Meta removed the separate replays metric when it introduced Views, and TikTok counts plays separately per impression and excludes replays. Retensis suggests plays divided by unique viewers as the figure that captures replay behaviour. Treat it as a derived estimate and keep it out of the completion rate.
How YouTube reports completion
YouTube separates the public view count from the numbers that pay creators. Public views are counted the moment a video starts to play across Shorts, long-form videos and live streams. Partner Program earnings are based on engaged views and engaged watch hours instead, and eligibility uses qualified views. A public view is therefore a reach number, not a completion number.
To read completion, use the two analytics views that describe the shape of the watch: average percentage viewed and the retention graph. The graph shows how much of the audience is still present at each second, so the height at the final second is the completion share. Hootsuite calls YouTube's version average percentage viewed rather than a binary completion, which is why a single pass or fail line is the wrong way to report it.
Public view counts are also cleaned and adjusted. YouTube says it may slow down, freeze or change a metric count and discard low-quality playbacks while it verifies them, and Realtime activity is only an estimate that may not match the watch page. Since November 2026 the public count on watch pages also includes an estimate for people watching together on TV screens. Record the observation date beside every YouTube rate so a later reader can tell a real change from a counting change.
Length matters because YouTube now treats longer vertical video as Shorts. YouTube allows Shorts up to three minutes, and a square or vertical video up to three minutes uploaded on or after October 15, 2024 is categorized as a Short. A completion rate for a 90-second Short and a 20-second Short are not comparable, so keep the length band attached to the number.
How Instagram and Facebook report completion
Meta moved to one distribution metric across its apps. Views are the number of times a reel or video was played, and they are calculated the same way on Facebook and Instagram. Views also capture when people look at the same content multiple times. That makes Views a useful reach number and a poor completion number, because a repeat look and a first look land in the same count.
The completion field is missing by design. Meta removed the separate replays metric when it introduced Views, and its watch-time updates describe minutes, not a finished-play count. Watch Time is the total time the reel or video was played, including time spent replaying. A completion estimate has to come from the watch-to-end data the dashboard shows, and it should carry a note that it is an estimate.
This is where platform guides can mislead. Hootsuite notes that TikTok reports completion as a share of viewers who watched in full, while YouTube frames the same idea as average percentage viewed. Instagram sits closer to YouTube's side of that split: you can read a shape of the watch, but you cannot lift a ready-made completion percentage. Do not copy a LinkedIn or TikTok completion rule onto Meta and call it native.
The practical Meta routine is short. Export the reel or video's views, watch time and any watch-to-end detail the account can see, compute the completion share from that export, and tag the row with the calculation date. Because Views include repeat looks, a Meta completion estimate will usually read slightly lower than a start-based rate on a platform that excludes replays.
How TikTok reports completion
TikTok Ads Manager has the most explicit vocabulary of the three. A 2-second video view is playback of at least two seconds within an impression session, a 6-second video view needs at least six seconds or a full play of a shorter video, and video views are the number of times the video started to play. Plays are counted separately for each impression and replays are excluded.
Completion has its own ladder. Video views at 25, 50, 75 and 100 percent count the plays that reached each share of the video length. The 100 percent line is the numerator, and video views are the denominator. The lower buckets are the diagnostic: a large gap between 50 and 75 percent says viewers leave in the middle, while a large gap between 75 and 100 percent says the ending loses them.
The 6-second focused view is a separate early-read signal. TikTok Ads says a low 6-second focused view count means people are dropping off early in the video. Read it beside the 25 percent bucket, because both answer the hook question, not the ending question.
One scope warning travels with all of these definitions: they are paid-reporting terms in TikTok Ads Manager, not organic Analytics labels. Use them to measure a paid placement precisely, and use the organic dashboard's own labels for an organic post. TikTok's average play time per user includes time spent replaying the video, so a high average play time with a modest completion rate is possible and is not a contradiction.
Original diagram of what each platform counts and what you can read as a completion. Sources: YouTube Help, Meta for Creators and TikTok Ads Manager, read November 17, 2026.
Calculate a comparable completion rate
The formula is short, and the discipline is in the labels. Hootsuite calculates completion rate by dividing completed views by total views and multiplying by 100, and its own metric line divides completions by video impressions. Both are valid if you say which one you used. Pick one denominator per platform and keep it fixed across every run.
- Choose the platform's own start count as the denominator: public views on YouTube, Views on Meta, video views on TikTok Ads. Write the exact metric name in the row.
- Choose the end count: the final second of the YouTube retention graph, the watch-to-end count you can export on Meta, or video views at 100 percent on TikTok Ads.
- Divide the end count by the start count and multiply by 100.
- Record the video length, the observation date and the metric definitions beside the rate.
- Record a rewatch estimate separately, using plays divided by unique viewers where both numbers exist.
| Platform | Numerator to use | Denominator to use | What to say about it |
|---|---|---|---|
| YouTube | Viewers still present at the final second of the retention graph | Public views | A read from the graph, not a separate published completion metric |
| Instagram and Facebook | Watch-to-end count exported from the dashboard | Meta Views | An estimate, because Views include repeat looks and replays were removed |
| TikTok Ads | Video views at 100 percent | Video views | A native paid-reporting metric, with replays excluded per impression |
Worked example, with invented numbers: a video logged 1,000 starts and 250 plays that reached the final second. The completion rate is 250 divided by 1,000, or 25 percent. If the same account's last five videos at the same length averaged 31 percent, this one is below its own baseline, and that is the comparison worth acting on. The invented figures are here to show the arithmetic, not to suggest a target.
Read completion against length and average view duration
A completion rate without its runtime is hard to interpret, because every extra second is another chance for a viewer to leave. Retensis bands its 2026 advice by length: above 40 percent under 15 seconds, above 30 percent at 15 to 30 seconds, and above 25 percent at 30 to 60 seconds. Read those as a starting band from one vendor, not as a platform rule, and compare your video with your own videos in the same band first.
The gap between average view duration and completion diagnoses the fault. Retensis says a Short can show a healthy average view duration and a poor completion rate at the same time, which points to an ending problem. Moderate average duration with high completion points to a self-selecting audience: the opening filters, and the viewers who stay finish. Both low points to a hook problem, where nothing later can recover the viewers who already left.
Looping complicates the read. Retensis notes that completion feeds looping, because a viewer who reaches the end sees the Short restart and some keep watching. That extra watch time is real, but it is not new viewers, so keep the rewatch estimate separate from the completion rate. A rise in both is a stronger signal than a rise in watch time alone.
Finally, put the video beside accounts of a similar size. Metricool analysed 375,462 accounts across Reels, TikTok and YouTube Shorts and advises comparing performance with accounts of similar size instead of chasing viral numbers. A 1,000-view video can be a strong result for a small account and a weak one for a large account, so the account-size context belongs in the row.
Original diagram of length bands and diagnostic cases. Length bands and cases follow Retensis, read November 17, 2026, and are labelled vendor advice, not a platform rule.
Build a weekly video scorecard
A weekly scorecard turns the definitions into a routine. One row per video, one fixed set of fields, and one review that produces a single edit decision for the next week. The row is the evidence; a dashboard percentage is only a summary of rows.
| Field | What it records | Why it belongs |
|---|---|---|
| Date and platform | When the video was read and where | The definition of a view changes by platform and over time |
| Length | Runtime in seconds | Completion targets move with length |
| Starts | The platform's own view count | It is the denominator you chose |
| End count | Final-second viewers, watch-to-end count or 100 percent views | It is the numerator you chose |
| Completion rate | End count divided by starts, times 100 | The headline number, with its definition attached |
| Average view duration | The middle of the distribution | It separates an ending problem from a hook problem |
| Rewatch estimate | Plays divided by unique viewers where available | Keeps looping out of the completion rate |
- Export the week's videos into the row shape above, one row per video.
- Group the rows by length band before you read any average.
- Read a run of five to ten videos, not a single video, because one result is noise.
- Compare each row with the account's own recent baseline and with accounts of a similar size.
- Write one edit decision for the next week, such as trimming the ending, tightening the middle or testing a shorter length.
Keep the routine small enough to survive a busy week. Hootsuite's advice is to pick two or three metrics that match the goal and ignore the rest, and completion belongs in that short list when the goal is attention. On autoplay feeds such as TikTok, Reels and Shorts, play rate loses meaning because the video starts on its own, so completion and retention carry the read there.
What completion cannot tell you
Completion measures attention, not results. AppsFlyer notes that a completed viewing still does not prove a conversion, and recommends follow-up metrics such as click-throughs or brand recall. It also warns that completion rate does not tell you how long most viewers watched or whether they paused or replayed any sections. A high rate with no click can be a well-made video with a weak ask.
Completion does not prove distribution either. No platform publishes the arithmetic behind its recommendation system, and Retensis explicitly benchmarks against targets rather than against a trigger point. Read a run of videos over several weeks before claiming that a change in completion changed reach. One video below a target means very little.
The measurement itself has to stay yours. Superdirector does not read watch time, completion or retention from a platform dashboard. It is a social publishing and research workspace that tracks accounts, competitors and keywords, makes carousels, X posts and video scripts, and publishes an account's eligible Instagram carousels and single X text posts once account-level posting is turned on. The publishing boundary is stated in Superdirector's llms.txt, and the scorecard in this guide stays with your own analytics.