Grow Creator Field Notes

AVD vs APV on YouTube Shorts: which matters

For Shorts, average percentage viewed is the number to act on. AVD only means something when you compare Shorts of the same length. What Studio shows.

For Shorts, average percentage viewed is the number to act on. YouTube calculates average view duration and average percentage viewed from the same pool of viewers, but AVD climbs with video length while APV is already scaled to it, so APV is the one that survives a comparison between two Shorts. AVD earns its keep in one situation: comparing Shorts of the same length.

That is the answer. The rest of this page is the part the comparison usually skips. What each number is measured against, why APV can read above 100%, what the March 2025 view-count change did to both, and a set of checks you can run this afternoon against numbers already sitting in YouTube Studio.

Both numbers are scored on the same filtered audience

YouTube's help documentation defines average view duration as the average minutes watched, and average percentage viewed as the average share of the video watched. Both definitions end with the same qualifier: "among those who stayed to watch". For Shorts, both are calculated from engaged views and the watch time attached to them, and all of that sits on YouTube's content performance help page. Engaged views are the ones YouTube describes as the viewers who chose to continue watching. The watch-time threshold that separates a view from an engaged view is not published, so nobody outside YouTube can tell you where the line falls.

So neither number tells you anything about the people who swiped away immediately. Both describe the audience that already opted in. If a Short reached 40,000 people and 3,000 of them kept watching, APV grades those 3,000.

A healthy APV is therefore a narrow claim. It says the video works for the slice that chose to keep watching, and says nothing about the people who never did.

Side by side

One caveat on that comparability row. Dividing by length strips out the arithmetic bias, not the behavioural one: longer Shorts still tend to post lower percentages, which is why every check below groups by length band before comparing anything.

Studio moves things around between releases, so if a field is not where the table says, search the metric name in Studio's own help rather than trusting any screenshot.

The length trap, with the arithmetic

Work the arithmetic on two Shorts of different lengths.

• A 15-second Short at 90% APV gives about 13.5 seconds of average view duration.
• A 55-second Short at 40% APV gives about 22 seconds.

The second one has an AVD roughly 60% higher and far worse retention. Sort your Shorts by AVD and you have mostly sorted them by length. That is the whole reason this comparison exists, and it is why the honest instruction is to hold length constant or ignore the number. YouTube makes the same point in a broader form in its Content tab analytics tips, where it tells creators to compare videos of the same type because audience behaviour differs across formats.

What YouTube does not publish is how heavily either number weighs in recommendation. Anyone quoting you a specific ranking weight for APV is guessing. Treat both as diagnostics for your own edits rather than as a model of the system. If the length question is the live one for you, our note on how long a Short should be for retention works through the trade-off.

APV above 100% is not a bug

This is the most misread number in Shorts analytics, and two documented facts explain it without any guesswork.

The first is the one above: APV is watch time measured against engaged views. The second is that YouTube's audience retention documentation notes that within a single view the same viewer can watch portions of a video more than once, which is why segment-level view counts sometimes exceed a video's overall view count.

Put those together and the arithmetic falls out. The watch time attached to a single engaged view can exceed the length of the Short, so the average percentage lands above 100%. Rewatching is the reading that fits, and it is the one the documentation supports. Studio does not surface a loop rate, so APV above 100% is the nearest thing to a rewatch signal you get.

How far above 100% counts as good? YouTube publishes no benchmark for this, and we are not going to invent one. The comparison that works is your own median inside a length band. We wrote up the honest version of that exercise in Shorts retention benchmarks.

What the March 2025 view change did to both numbers

From 31 March 2025, YouTube changed how Shorts views are counted: views now count each time a Short starts to play or replay, with no minimum watch time. The old metric was kept and renamed engaged views, and YouTube states that Partner Program eligibility and Shorts revenue sharing still run on engaged views. That change is documented on YouTube's Shorts help page.

Three consequences for this comparison.

1. Your view count jumped. Your AVD and APV did not, because both still run on engaged views.
2. Do not calculate your own AVD by dividing watch time by views. That mixes two different populations and gives you a number well below the one Studio shows.
3. The gap between views and engaged views became useful on its own. It is the closest thing Studio gives you to a hook score: of the people who started the Short, how many chose to keep watching.

Six checks against numbers you already have

Run these in order. Every one uses data already in Studio.

1. Bucket first. Split your last 20 to 30 Shorts into three length groups, roughly under 20 seconds, 20 to 40 seconds, and over 40 seconds. Do not compare across groups yet.
2. Find your own APV median inside each group. That median is your benchmark. Any figure from a blog post, ours included, is a worse reference than your own last month.
3. Check engaged-view share on your best and worst Shorts. Divide engaged views by views. If it is low everywhere, including on your high-APV Shorts, the bottleneck is the opening frames rather than the edit.
4. Now compare AVD inside one group. With length held constant, an AVD gap is a real retention gap, and it arrives in seconds, which is easier to act on than a percentage. Six seconds against a group median of eleven gives you a timestamp to scrub to.
5. Open the retention graph on the weakest Short in your strongest group and mark the first steep drop. That is where the cut goes.
6. Find every Short with APV above 100%. Whatever those share, whether a loop-friendly ending, dense information, or a punchline that sends you back to the start, is your template.

Step three is where the answer usually turns up, and it is the step that gets skipped. If you would rather have that read done for you, the free YouTube channel audit runs on a public channel with no signup and points at whichever gate is capping it. For a per-video read of hook, retention and rewatch on a single Short, Reel IQ does the same job one video at a time.

Three questions, three different fixes

• Engaged-view share well below your own median? Fix the hook. The cover frame, the first spoken clause and the first legible words on screen decide this, and an edit deeper into the video cannot move it, because the viewer has already decided by then.
• Engaged-view share normal, APV below your median for that length group? Fix the middle. Open the retention graph, find the first cliff, cut it.
• APV strong, AVD low against your top performers, engaged-view share fine? Fix the length. You have a structure that holds attention over a short run, so test the same structure with more in it, or build an ending that feeds back into the opening.

For the metric-by-metric read rather than this one comparison, there is a longer walkthrough on reading your Shorts analytics.

When AVD genuinely matters for Shorts

Plenty of pages on this topic say "use APV" and stop, which leaves you without a rule for the cases where seconds are the better unit. There are three.

• You publish a fixed-format series at a consistent length. Length is already held constant, so AVD in seconds is the cleaner edit target.
• You are testing two cuts of the same video. Same length, same content, so seconds watched is the direct comparison.
• You care about total watch time rather than reach. Read watch time directly in Studio instead of multiplying AVD by views, since those two use different view populations.

Outside those, sort by APV.

The same split on Instagram Reels

Instagram gives you the seconds and leaves the percentage to you. Per-reel insights report an average watch time with no APV sitting next to it, so if you want the percentage you divide average watch time by the length of the reel yourself. The fields live in Instagram's reel insights.

The denominator is the thing to check before you compare anything. YouTube states its denominator plainly: engaged views, so the people who stayed to watch. Instagram does not use that concept, and it renames and re-bases these fields between app versions, so read the definition Instagram shows beside the metric in your own app rather than assuming it lines up with YouTube's. Until you have done that, treat a Reels-seconds against Shorts-seconds comparison as unsafe.

What to do tomorrow

Open Studio, filter to Shorts, and group your last 30 by length instead of by views. Inside the biggest group, note the median APV and the engaged-view share. If engaged-view share is the weaker of the two, your next three Shorts should change nothing but the first two seconds: cover frame, opening line, on-screen text. If APV is the weaker one, keep the openings and cut roughly a third out of the middles. Check again in two weeks against the same length group, not against anyone's benchmark.

Canonical: https://growcreator.pro/blog/avd-vs-apv-youtube-shorts

Leumos Private Limited · hello@growcreator.pro