# Reels analytics: read your numbers, then fix or repeat

What each number in your Instagram reel insights means, the four worth tracking, and a weekly habit that tells you to kill, recut or share every reel.

Source page: https://postmosis.com/vault/reel-insights. By Leo at Postmosis. Last checked 2026-10-29. Adapted from Keepy Labs: [Short-Form Video Master Course: Read your numbers](https://keepylabs.com/resources/short-form-video-course/read-your-numbers), [Short-Form Video Master Course: Edit for retention](https://keepylabs.com/resources/short-form-video-course/edit-for-retention), [Check Your Reel Before You Post](https://keepylabs.com/resources/claude-virality-predictor).

In short: To see a reel's analytics, open it from the Reels tab on your profile and tap View insights [1]. Read four things: skip rate for the hook, average watch time for the body, the retention graph for the second people leave, and shares, saves and comments for whether the right people cared [2]. Compare each reel with your own median, then kill it, recut it or make more like it.

## Ten minutes on a Friday

You post a reel on Tuesday, between a client call and an invoice. By Friday you've glanced at the view count twice and still can't say whether it worked. A habit of about 10 minutes a week fixes that [2]:

1. Open every reel that's been up for 48 to 72 hours and give it a row in the tracking sheet below [2].
2. Copy four things into each row: skip rate, average watch time, the second people drop off, and shares, saves and comments.
3. Put each one next to your own median and mark it kill, recut or share.
4. Read the comments on the share reels, and pick next week's topic from what people asked.

Expect about half your reels to land below your median, because that's what a median is. The habit is there to catch the few that clearly beat it, and to stop you making more of the ones that quietly failed.

## Where to find reel insights

1. Tap your profile picture to go to your profile [1].
2. Open Reels and tap the reel you want to check [1].
3. Tap View insights.

You can also see reels in your account insights [1], with any time frame within the past 90 days [3]. Tap the (i) in the top right corner for Instagram's definition of each metric [3].

You need a public account, personal or professional [1]. Insights cover reels posted on or after 14 May 2021 and stay available for two years after a reel goes up [1], so copy your numbers into a sheet if you want a longer record.

## What each number means

Instagram reel insights, defined:

| | What Instagram counts | What it tells you |
| --- | --- | --- |
| Views | Every time your reel starts to play or replay | How often it was played, which on its own books no calls [1] |
| Viewers | Unique accounts that saw your reel on screen at least once, played or not | How many people it reached [1] |
| Watch time | Total time your reel was played, replays included | Total attention across a run of reels [1] |
| Average watch time | Watch time divided by initial views, where an initial view is the first play in a session | Whether the body holds people [1][2] |
| Skip rate | The percentage of views from people who skipped in the first 3 seconds | Whether the hook worked [4][2] |
| Interactions | Likes, comments, saves and shares | Whether the right people cared [3][2] |
| Follows | Accounts that followed you after watching this reel | Whether people wanted to hear from you again [1] |

Viewers and skip rate are estimates that Meta still marks as in development [1] [4], so read them as a direction more than an exact count. Views and viewers differ too: one person who watches three times counts as three views and one viewer [1].

Likes are the weakest signal on the list. Saves, shares and comments with a question in them come closer to someone thinking about hiring you. The [vanity metrics](https://postmosis.com/glossary/vanity-metrics) and [reach vs impressions](https://postmosis.com/glossary/reach-vs-impressions) entries go further, and the [engagement rate calculator](https://postmosis.com/tools/engagement-rate-calculator) turns interactions into a rate you can compare.

## The four numbers to track

- Skip rate: how many people swiped away in the first seconds. It tells you if the hook worked [2].
- Average watch time: how long people stayed. It tells you if the body worked [2].
- The retention graph: where exactly people left. It tells you which second to fix [2].
- Shares, saves and comments with your keyword: whether the right people cared, which is where enquiries come from [2].

If a reel's insights have no retention graph, compare its average watch time with its length. The bigger the gap, the earlier people are leaving, and the more the opening needs work.

## Read the retention graph

The retention graph shows when you lose your audience during the reel [5]. It plots how much of the audience is still watching at each moment. YouTube's retention report reads the same way, and its intro figure is the percentage of your audience still watching after the first 30 seconds [6].

Example: 90% retention at second 3 means 9 in 10 people who started your reel were still there at second 3, so the hook held. 40% at second 20 means 4 in 10 stayed that long. Neither number is good or bad on its own. Put it next to your other reels of a similar length.

The line always slopes down, and the flatter it is, the more engaged your audience is [5]. Read its shape before you read any single number:

- A flat stretch: people are watching that part from start to finish [6].
- A gradual decline: interest fading over time, which nearly every video does [6].
- A dip: people leaving or skipping at that moment [6]. That's the second you fix.
- A spike: more people watched that part again or shared it [6]. Whatever is on screen there, do more of it.

Where the biggest dip sits tells you what to change. A dip in the first 3 seconds means the hook needs swapping. One in the middle usually points to a slow beat you can cut, and one just before your call to action means people got what they came for, so move the call to action earlier [7].

## Judge against your own median

Whether 5K views is good depends entirely on the account [2]. Benchmarks from other accounts pull you toward the wrong things. Compare every reel with your own median and the winners stand out, even on a small account [2].

Your median is the middle number, and one big hit can't drag it up the way it drags an average. To find it:

1. List the views for your last 30 reels [2].
2. Sort them from lowest to highest.
3. Take the middle value. With an even count, average the two in the middle.

Do the same for skip rate and average watch time. Then every new reel gets a simple test: better or worse than your usual?

Example: Say your last ten reels got 180, 220, 240, 260, 300, 310, 350, 400, 900 and 2,400 views. The two in the middle are 300 and 310, so your median is 305. The average is 556, dragged up by the one hit. A new reel at 650 views is more than twice your median, so its topic is one to make more of. Next to the average it would have looked ordinary.

That's also the honest answer to 'what's a good skip rate?' or 'what's a good retention percentage?'. Whatever beats your own median is good, because what counts as good depends entirely on the account [2].

Once a month, run it again on your last 30 reels. Anything at 2x your median or more is a topic to make more of [2].

## Kill, recut or share

Decide 48 to 72 hours after a reel goes up [2]. Any earlier and you're judging a reel Instagram may still be testing on its first small audience [8].

What to do with each reel:

| | The signs | What to do next |
| --- | --- | --- |
| Share and make more | 2x your median views or more, or well above your usual shares and saves | Treat it as a winning topic and plan more videos on it [2] |
| Recut | A strong hook (low skip rate), but people leave in the middle | Fix the drop-off second and test again [2] |
| Kill | A high skip rate and nothing above your usual | Move on, and don't keep reposting it [2] |

Example: A business coach checks three reels on Thursday, 72 hours after posting the first. 'Why your first hire should be part-time' has 2.5 times her median views and six comments asking about contracts: share and make more, and the first new one is about contracts. 'Pricing your first package' has her best skip rate in a month, but watch time falls away at second 11, where she lists five options: recut it with the list on screen. 'My morning routine' has a high skip rate and nothing above her usual: kill it, and don't repost.

Read the comments on every reel you mark Share before you plan more [2]. They tell you which part of the topic people want next. For a recut, the [trial reels loop](https://postmosis.com/vault/instagram-trial-reels) tests your fix on strangers first. For a winner, [repurpose content](https://postmosis.com/vault/repurpose-content) turns it into ten more videos.

Prompt: sort my reels:

```
Here's my tracking sheet for my last [n] reels: [paste]. My median views: [n].
Mark each reel Kill, Recut or Share with these rules:
Share: 2x my median views or more, or clearly above my usual shares and saves.
Recut: skip rate better than my average, but watch time drops mid-video.
Kill: everything else.
Then list the hook types and topics that show up most among the Share reels, and the 3 reels I should make next.
```

## When the numbers mislead

- Too few reels: with fewer than ten, your median moves every time you post. Until then, judge by comments and enquiries, and keep posting.
- One lucky share: a reel can take off because one big account shared it. Check who's commenting. If they're not people you could work with, the topic hasn't proved itself yet.
- A seasonal topic: a reel about a tax deadline or the school holidays beats your median the week before and flops a month later. Compare it with the same week last year.
- A reel that's still being tested: before 48 hours, Instagram may still be showing it to its first small audience [8], so wait until 48 to 72 hours to decide [2].

## What insights leave out

Instagram's insights tell you what happened to each reel. They don't tell you what to make next, and they miss the things that matter most when your posts are meant to bring in clients:

- Who watched: a view from a stranger on the other side of the world counts the same as one from someone in your area who might hire you.
- Which reel brought the enquiry: a DM that says 'can we talk?' or a call booked from your link in bio doesn't show up next to the reel that caused it.
- Patterns across reels: insights won't tell you that your contrarian hooks beat your questions, though a hook-type column in your sheet will.
- How far back you can look: account insights cover time frames within the past 90 days [3], and a reel's own insights stay up for two years [1].

Paid analytics tools fill in some of this. If you post a few reels a week, a sheet and about 10 minutes a week cover most of it [2], as long as you add one column the apps can't fill: enquiries.

## Log every reel in one sheet

Give it about 10 minutes a week [2] and log each reel within 72 hours of posting it [2]. Use one row per reel, or one per version when you're testing:

Reel tracking sheet:

```
Date | Reel | Hook type | Views | Skip rate | Avg watch time | Drop-off second | Shares + saves | Decision
 | | | | | | | | Kill / Recut / Share
```

Postmosis audits your public Instagram or TikTok from its last 30 posts and tells you what to double down on, fix or stop, with your best posts to remix. [See the account audit](https://postmosis.com/features/social-media-audit)

## Find the flat spots

People rarely leave at a big moment. They drift during a flat spot: two or more seconds where nothing new happens [9]. Find those seconds and a cut or a zoom usually fixes them.

After you post, the retention graph shows where they are. Before you post, you have two ways to find them:

- Watch it like a stranger. Play the reel muted on your phone and write down the timestamp each time two seconds pass with nothing new on screen [9].
- Use an attention predictor: some AI tools score a short clip and draw a predicted attention curve, second by second [10]. It's a prediction from an AI model with no real viewers behind it [10], so treat it as a second pair of eyes.

With a predictor, score the hook on its own first. Run the first 15 seconds and look at how far the curve dips after second 1, and whether it recovers by second 3 [9]. Every clip dips a little at the start, and a small dip that recovers fast is fine [10].

For a longer reel, split it into 15-second pieces that overlap by 1 second, and mark every stretch where the curve stays flat for 2 seconds or more [9]. A flat spot over your end card or call to action is normal, so ignore it [10].

A predictor only sees the video. It can't judge your caption, your audio, your cover or when you post [10].

## Fix each flat spot

Flat spots and their fixes:

| | The fix |
| --- | --- |
| The same shot held for 3 seconds or more | Cut it, or zoom in by 10 to 15 percent [9] |
| A card with only text | Show the thing moving instead [9] |
| Half the frame empty | Fill it with proof, a screen recording or an object [9] |
| Just your face, with nothing happening | Go full frame on the proof, or cut to it [9] |
| A long sentence with no payoff on screen | Put the payoff word on screen as you say it [9] |

Then check the fixed version and compare the curves. You want a curve with no long flat stretches, and a higher score on its own isn't the goal [9]. Look at where the flat spots moved more than at the score itself [10].

Prompt: fix one flat spot:

```
I checked my reel for attention. Here are the frames from a stretch where attention goes flat.
Reel topic: [one line]
Flat spot: seconds [start] to [end]
What happens after second 1: [describe the dip]
Look at each frame and tell me:
1. What's on screen, in plain words.
2. Why attention probably drops here. Pick from what the frames show: the same shot held too long, a text-only card, an empty half of the frame, a face with nothing else moving, a repeated layout, a long sentence with no payoff word on screen. If nothing looks wrong, say so. Don't invent a reason.
3. One edit, with the exact second it applies to. Good: "At 23 to 25 seconds, swap the second talking head for the screenshot you're describing, full screen." Bad: "make it more engaging."
Keep it short. One edit per flat spot.
```

The Postmosis video editor cuts the pauses, retakes and filler words in the video you filmed, adds zooms, puts your hook on screen, adds captions in 6 styles and places B-roll for you. [See the video editor](https://postmosis.com/features/video-editor)

## Check the first three seconds

Fix every no on this list before you post. If the opening needs rewriting, the [video hooks](https://postmosis.com/vault/video-hooks) playbook has the formulas.

Hook checklist:

```
[ ] Is there a number or a clear result in the first 5 words?
[ ] Is the hook 18 words or fewer?
[ ] Is something moving on screen in the very first frame?
[ ] Is there a 2 to 6 word headline on screen in the first second?
[ ] Does the first spoken word appear on screen within a quarter of a second?
[ ] Does it open on the best moment instead of building up to it?
[ ] Can people see the thing you're talking about?
[ ] Does something new appear by second 2?
[ ] Does your face share the frame with something else?
[ ] Is the payoff word on screen when you say it?
```

## Before you post

Retention checklist:

```
[ ] No shot is held longer than 3 seconds without something changing.
[ ] Something new appears by second 2.
[ ] Every number or main word you say is on screen when you say it.
[ ] Captions are on, sized for a phone, inside the safe area.
[ ] No flat spot of 2 seconds or more is left in the final cut.
[ ] The last second makes sense if the video loops back to the start.
```

Example: A typical talking-head reel: the hook holds, then the curve goes flat from second 4 to second 7 while you explain the setup with nothing new on screen. Two fixes: a zoom of 10 to 15 percent on the main sentence, and the one number from that sentence on screen, both of which take seconds in any editor. Check it again and the flat stretch is gone.

Instagram can add closed captions for you with speech recognition [11]. Keep most reels under 3 minutes too, because longer ones won't be recommended to new audiences [12].

## Common questions

### How do I see analytics for my Instagram reels?

Go to your profile, open Reels, tap the reel and tap View insights [1]. You need a public account. For trends across many reels, use your account insights, which cover any time frame within the past 90 days [3].

### What is a good skip rate on Instagram reels?

There's no single good number. Skip rate depends on your topic, your audience and who the reel was shown to. Compare each reel with your own median instead: lower than your usual means the hook worked [2].

### What's a good audience retention percentage?

No fixed percentage works as a target, because it depends on the reel's length and your account. A flatter curve means more people stayed. Compare a reel's curve with your other reels of a similar length, and fix the steepest drop first [7].

### What's the difference between views and viewers?

Views count every time your reel starts to play or replay. Viewers count the unique accounts that saw it on screen at least once, played or not [1]. One person who watches three times is three views and one viewer.

### How is average watch time calculated?

Instagram divides total watch time by the number of initial views. An initial view is the first time your reel plays in a reel session, so replays add to watch time without adding to the count [1].

### How long should I wait before judging a reel?

Two to three days. Decide at 48 to 72 hours [2]. Instagram shows a reel to a small audience first and widens it for the best performers [8], so an early read can mislead you.

### Why did my reel stop getting views?

Usually because it stopped winning the next round. Instagram shows eligible reels to a small group, then shows the top performers to a wider group, and so on [8]. Reels over 3 minutes also aren't recommended to new audiences [12].

## Sources

1. [View insights on your Instagram reels, Instagram Help Centre](https://help.instagram.com/202865988324236), read 2026-10-29
2. [Read your numbers: kill, recut or share, Keepy Labs](https://keepylabs.com/resources/short-form-video-course/read-your-numbers), read 2026-10-29
3. [About Instagram insights, Instagram Help Centre](https://help.instagram.com/788388387972460), read 2026-10-29
4. [Instagram Media Insights, Meta for Developers](https://developers.facebook.com/docs/instagram-platform/reference/instagram-media/insights), read 2026-10-29
5. [Instagram Adds Retention Metrics for Reels, Social Media Today](https://www.socialmediatoday.com/news/instagram-adds-retention-insights-reels/758464/), read 2026-10-29
6. [Measure key moments for audience retention, YouTube Help](https://support.google.com/youtube/answer/9314415?hl=en), read 2026-10-29
7. [The 48-Hour Trial Loop, Keepy Labs](https://keepylabs.com/resources/trial-loop), read 2026-10-29
8. [Helping creators find new audiences, Instagram for Creators](https://creators.instagram.com/recommendations-and-Originality), read 2026-10-29
9. [Edit for retention: find and fix flat spots, Keepy Labs](https://keepylabs.com/resources/short-form-video-course/edit-for-retention), read 2026-10-29
10. [Check your Reel before you post it, Keepy Labs](https://keepylabs.com/resources/claude-virality-predictor), read 2026-10-29
11. [Manage closed captions for your reels on Instagram, Instagram Help Centre](https://help.instagram.com/7487270478066359), read 2026-10-29
12. [Record a reel on Instagram, Instagram Help Centre](https://help.instagram.com/225190788256708), read 2026-10-29