Which YouTube Competitor Metric Should You Trust? We Analyzed 2,971 Videos

Open a competitor channel and ask a simple question:

Which video should I study?

You might think the answer is obvious.

Pick the video with the most views.

But change the metric and you can get a completely different winner.

So we tested it.

OverseerOS ranked the same 2,971 long-form YouTube videos across 107 channels and 13 niches four different ways:

1. Total views
2. Age-adjusted views per day
3. Views per subscriber
4. Channel-relative outlier score

The result was stronger than we expected.

Across all 13 niches, not a single niche had the same #1 video under all four metrics.

Raw views and views per subscriber chose the same #1 video in:

0 of 13 niches.

Raw views and outlier score chose the same #1 video in:

1 of 13 niches.

Even when we expanded the comparison from the single winner to the top 10 videos, raw views and outlier score shared only:

2.3 videos out of 10 on average.

That means roughly three-quarters of the videos you would prioritize by raw views disappeared when the question changed to:

Which video most dramatically beat what was normal for its own channel?

This is why YouTube competitor analysis goes wrong so easily.

The metrics are not interchangeable.

They answer different questions.

And if your goal is finding competitor videos worth reverse-engineering for your next idea, our data suggests a clear default:

Start with channel-relative outlier performance. Then add recency, absolute reach and channel-size context. Do not start with raw views alone.

Key Findings

The most important finding is simple:

There is no universal "best-performing competitor video." There is only the best video for the question you are trying to answer.

What Is the Best YouTube Competitor Metric?

If your goal is finding ideas worth studying, channel-relative outlier score is the strongest default starting point of the four metrics we tested.

But each metric has a legitimate job.

The mistake is not using any of these metrics.

The mistake is asking one metric to answer a question it was not designed to answer.

How We Analyzed the Data

This study used public YouTube performance information observed and analyzed by OverseerOS.

The research was frozen on September 13, 2026.

The central research question was:

If a creator uses different public performance metrics to choose which competitor videos deserve attention, how often do those metrics actually identify the same videos?

The final sample

A video had to meet all of these conditions:

• Long-form
• Between 30 and 365 days old
• Usable public view data
• A usable public subscriber count for the channel
• A high-confidence OverseerOS niche classification
• At least 20 earlier long-form videos available from the same channel

We then required each included niche to contain:

• At least 100 qualifying videos
• At least 5 separate channels

That produced:

• 2,971 videos
• 107 channels
• 13 niches

The niches were:

• Gaming
• History
• AI
• Top lists
• Storytime
• Psychology
• Technology
• Education
• Finance
• Spirituality
• News
• Animation
• Philosophy

The median qualifying video had:

• 86,859 public views
• 101 days of age
• A channel with approximately 362,000 subscribers

Metric 1: Total Views

This is the simplest metric.

Total views = current recorded public video views

It answers:

Which videos accumulated the most views?

That is useful.

But it does not adjust for:

• Channel size
• Video age
• The channel's normal performance
• Whether 2 million views is exceptional or ordinary for that creator

Metric 2: Views Per Day

We calculated:

Views per day =
Current public views
÷
Video age in days

This is an age-adjusted average.

It is not true real-time views per hour or recent view velocity.

A video with:

1,000,000 views after 50 days

would have:

20,000 average views per day.

A video with:

1,000,000 views after 250 days

would have:

4,000 average views per day.

The first video therefore ranks higher.

This helps prevent older videos from automatically dominating every comparison.

For a dedicated analysis of recent momentum, see our study of YouTube view velocity vs total views.

Metric 3: Views Per Subscriber

We calculated:

Views per subscriber =
Video views
÷
Current channel subscriber snapshot

Example:

A video has:

500,000 views

on a channel with:

100,000 subscribers

The ratio is:

5.0 views per subscriber.

A 500,000-view video on a 5-million-subscriber channel would produce:

0.10 views per subscriber.

This metric therefore rewards videos that reach far beyond the apparent size of their channel.

We have analyzed this metric separately in our YouTube views-to-subscriber ratio study.

Metric 4: Channel-Relative Outlier Score

For this study, the outlier baseline was the median performance of the video's previous 20 long-form uploads.

We calculated:

Outlier score =
Video views
÷
Median views of the previous 20 long-form videos

If a channel's previous 20 videos had a median of:

40,000 views

and the new video reached:

400,000 views

its score would be:

10x baseline.

This answers a fundamentally different question from raw views:

How abnormal was this result for this channel?

That distinction turns out to matter enormously.

Finding 1: The Four Metrics Never Agreed on the Same Winner

For every niche, we ranked every qualifying video four times.

One ranking by:

• Raw views
• Views per day
• Views per subscriber
• Outlier score

Then we compared the #1 video.

Across 13 niches:

Not one niche had the same #1 winner under all four methods.

That alone should change how creators think about competitor analysis.

The metric is not a cosmetic choice.

It changes what you study.

Pairwise winner agreement

Think about what this means.

A creator could research the exact same niche, use the exact same public dataset and reach a completely different conclusion simply because they clicked a different sort button.

That is not a small methodological detail.

It can determine:

• Which competitor you study
• Which topic you copy into your research board
• Which title you analyze
• Which thumbnail you reverse-engineer
• Which format you believe is winning
• Which video you spend thousands of dollars reproducing

Finding 2: Raw Views and Outlier Score Barely Agreed Even in the Top 10

Maybe comparing only the #1 video is too strict.

So we widened the test.

For each niche, we took the top 10 videos under each metric.

Then we measured how many videos appeared in both lists.

Raw views and views/day were the closest.

That makes sense.

Both still reward videos with large absolute view counts.

But raw views and outlier score shared only:

2.3 of their top 10 videos on average.

In other words:

About 77% of the videos selected by one of those methods were absent from the other's top 10.

That is not two slightly different rankings.

Those metrics are looking for different kinds of winners.

Finding 3: Raw Views Mostly Surfaced Big Channels

Now we can see why.

We examined the top 10 videos in every niche under each ranking method.

That created 130 selections per metric.

Here is what the typical selection looked like.

The raw-view ranking behaved exactly as its formula suggests.

It surfaced enormous videos from enormous channels.

The median raw-view top pick came from a channel with:

2.18 million subscribers.

But its median outlier score was only:

1.87x.

That is an important distinction.

A video can have millions of views and still teach you surprisingly little about what recently changed.

If the channel normally gets two million views, a three-million-view upload is successful.

It is not necessarily a strategic anomaly.

Why this matters for creators

Suppose Channel A normally gets:

3 million views

and one video reaches:

4 million.

Channel B normally gets:

25,000 views

and one video reaches:

300,000.

Raw views chooses Channel A.

Outlier analysis chooses Channel B.

If your question is:

Which creator has the bigger audience?

Channel A is relevant.

If your question is:

Which video contains a surprising signal I should investigate?

Channel B may be much more useful.

Finding 4: Views Per Day Fixed Age Bias but Still Favored Large Channels

Views per day produced a much younger set of winners.

The median age among its top selections was:

59 days.

Raw-view winners had a median age of:

180 days.

That is exactly why age adjustment can be useful.

But views/day did not suddenly transform the ranking into a small-channel discovery system.

Its typical top-ranked channel was even larger:

2.86 million subscribers.

The median views/day top pick had:

• 2.49 million views
• 21,447 average views per day
• 2.86 million subscribers
• Only 1.91x channel-relative performance

So views/day mostly answered:

Which large videos accumulated views quickly relative to their age?

That is useful for spotting freshness.

It is much less useful for answering:

Which competitor did something unusually successful relative to what they normally achieve?

Raw views and views/day were highly correlated

We ranked every video inside its niche and measured how closely the rankings moved together.

The average rank correlation between:

Raw views and views/day was 0.93.

That is extremely high.

For comparison:

So age adjustment mattered.

But views/day was still much closer to raw views than to channel-relative outlier performance.

This is an important warning for creators who treat average daily views as a complete normalization method.

It adjusts for time.

It does not adjust for channel expectations.

Finding 5: Views Per Subscriber Surfaced Much Smaller Channels

Views per subscriber produced a radically different competitor list.

Its median top-ranked channel had:

112,000 subscribers.

Compare that with:

2.18 million subscribers for raw-view winners.

The median views/subscriber winner had:

• 500,180 views
• 112,000 subscribers
• 4.36 views per subscriber
• 4.19x channel-relative performance

This makes views/subscriber useful for finding videos that punched above channel size.

It is especially useful when the creator question is:

Can a relatively small channel get serious reach with this kind of video?

That is a much more actionable question for many creators than:

What got the most views in the entire niche?

Views/subscriber and outlier score overlapped more

The average top-10 overlap between views/subscriber and outlier score was:

4.1 of 10.

That was substantially higher than the:

2.3 of 10

shared between raw views and outlier score.

Their average rank correlation was also:

0.61

versus only:

0.39

between raw views and outlier score.

That makes intuitive sense.

Both views/subscriber and outlier score attempt to provide context beyond absolute scale.

But they are still not the same metric.

Why Views Per Subscriber Can Still Mislead You

A high views-to-subscriber ratio does not automatically mean a video is an extraordinary outlier.

There are several reasons.

Subscribers are not the baseline

Imagine a channel with:

50,000 subscribers

but its last 20 videos all received:

250,000 views.

A new video gets:

300,000 views.

Its views/subscriber ratio is:

6x.

That sounds exceptional.

But relative to the channel's actual recent performance:

300,000
÷
250,000
=
1.2x baseline

The video is normal.

The subscriber denominator made it look extraordinary because subscribers were a poor representation of the channel's actual reach.

The subscriber snapshot changes over time

Our study uses the public subscriber snapshot available around the analysis period.

That is not necessarily the subscriber count the channel had on the exact day every historical video was published.

A channel may have grown substantially after a breakout.

That makes views/subscriber especially useful as a directional cross-channel metric, but weaker as a precise historical performance measure.

This is why we do not recommend treating it as a universal outlier score.

Finding 6: Outlier Score Found the Videos Raw Views Missed

The outlier-ranked top selections looked almost like a different YouTube ecosystem.

Their median numbers were:

• 288,746 views
• 122,000 subscribers
• 3,209 average views per day
• 1.96 views per subscriber
• 9.91x the channel's previous 20-video median

Compare that with raw-view picks:

• 3.40 million views
• 2.18 million subscribers
• 1.87x baseline

Raw views found bigger videos.

Outlier score found bigger departures from normal.

That is why outlier analysis is so valuable for competitor research.

It asks:

What changed?

rather than:

Who is already huge?

Outlier ranking also spread attention across more channels

For every niche, we measured how many distinct channels were represented among the top 10 videos.

Average channels represented:

Raw views often allowed dominant channels to occupy several of the top positions.

Outlier score spread the research pool across more creators.

That matters because studying only the biggest creator in a niche can produce a distorted view of what is actually transferable.

A channel with enormous authority can succeed with ideas that a smaller channel cannot.

Outliers can expose opportunities where the idea broke beyond the creator's normal distribution.

For a deeper explanation of the concept, see our guide to YouTube outlier analysis.

The Metrics Were Answering Four Different Questions

This is the cleanest way to interpret the study.

Raw Views

Question:

What achieved the most total reach?

Useful for:

• Understanding absolute market size
• Finding famous category-defining videos
• Seeing what topics have generated massive cumulative demand
• Studying dominant channels

Weak for:

• Finding hidden breakout opportunities
• Comparing small and large channels fairly
• Detecting unusual performance

Views Per Day

Question:

What accumulated views quickly for its age?

Useful for:

• Reducing age bias
• Prioritizing newer winners
• Finding relatively fresh momentum
• Comparing a 60-day-old video with a 300-day-old video more fairly

Weak for:

• Adjusting for channel size
• Determining whether performance is unusual for that creator
• Measuring true recent velocity

Views Per Subscriber

Question:

What reached far beyond the channel's current subscriber base?

Useful for:

• Finding small-channel reach
• Comparing differently sized channels
• Discovering accessible-looking breakouts
• Testing whether channel authority alone explains the result

Weak for:

• Channels whose normal views already greatly exceed subscribers
• Historical videos where subscriber count has changed
• Distinguishing 5x normal performance from 1.2x normal performance

Outlier Score

Question:

What most dramatically beat this channel's recent normal?

Useful for:

• Finding competitor anomalies
• Discovering breakout topics
• Identifying videos worth reverse-engineering
• Finding smaller channels with exceptional individual wins
• Separating channel scale from unusual performance

Weak for:

• Measuring absolute market size
• Measuring true current momentum
• Comparing channels without enough historical uploads
• Explaining why the video broke out

That last point matters.

An outlier tells you:

Something happened here.

It does not automatically tell you:

This topic caused it.

You still need to investigate.

Which Metric Should You Use for YouTube Competitor Research?

For most creators researching what to make next, we would use this hierarchy.

1. Start With Outlier Score

Find videos that dramatically beat the creator's normal baseline.

You are looking for:

Unexpected success

not merely:

Large creator + large view count

A 5x or 10x anomaly deserves investigation.

2. Add Recency or Velocity Context

A 15x outlier from four years ago and a 6x outlier happening now answer different questions.

Check:

• Publication date
• Views per day
• Recent velocity when available
• Whether the topic is still appearing among current winners

Do not treat old proof as automatically current proof.

3. Check Cross-Channel Topic Confirmation

One outlier proves:

This worked once.

It does not prove:

This topic transfers.

Look for independent success from other channels serving a similar audience.

Our separate analysis of whether one viral competitor is enough to validate a YouTube topic found that independent cross-channel confirmation materially strengthens the signal.

4. Check Raw Reach

Now ask:

How large has this demand actually become?

Raw views become useful again here.

A 20x outlier with 80,000 views and a 20x outlier with 8 million views are both strategically interesting.

But they may represent very different market sizes.

5. Use Views Per Subscriber as a Reality Check

Ask:

Could a smaller channel plausibly win here?

If smaller channels repeatedly achieve view counts well above their subscriber bases, the opportunity may be less dependent on existing audience scale.

This is especially useful when the biggest channels dominate the raw-view rankings.

The Best Competitor Research Stack Is Not One Metric

The strongest workflow is not:

Pick the perfect metric.

It is:

Use each metric to remove a different kind of bias.

A practical stack looks like this:

This is much stronger than sorting a competitor spreadsheet by one column and assuming the first result is the opportunity.

A Simple Example

Imagine three competitors.

Video A

• Channel subscribers: 3,000,000
• Video views: 4,000,000
• Channel normal: 2,500,000
• Age: 200 days

Metrics:

Raw views = 4,000,000
Views/day = 20,000
Views/subscriber = 1.33x
Outlier score = 1.6x

Video B

• Channel subscribers: 90,000
• Video views: 900,000
• Channel normal: 80,000
• Age: 120 days

Metrics:

Raw views = 900,000
Views/day = 7,500
Views/subscriber = 10x
Outlier score = 11.25x

Video C

• Channel subscribers: 2,000,000
• Video views: 2,000,000
• Channel normal: 1,500,000
• Age: 40 days

Metrics:

Raw views = 2,000,000
Views/day = 50,000
Views/subscriber = 1x
Outlier score = 1.33x

Now ask four questions.

Which had the most views?

Video A.

Which is moving fastest for its age?

Video C.

Which reached furthest relative to channel size?

Video B.

Which most dramatically broke its own channel pattern?

Video B.

There is no contradiction.

The metrics are simply answering different questions.

If you are looking for the most interesting competitor idea to investigate, Video B is probably where we would start.

How to Apply This With OverseerOS

The practical value of competitor research is not producing a prettier spreadsheet.

It is moving from:

Something worked.

to:

I understand which evidence matters, what I should investigate, and what I can adapt into an original video.

Step 1: Analyze the Channel

Use the OverseerOS YouTube Channel Analyzer to establish context around the competitor.

Look at:

• Normal performance
• Top videos
• Recent videos
• View distribution
• Breakout signals
• Upload behavior

The first job is determining whether the video you are excited about is actually unusual.

Step 2: Prioritize the Outliers

Do not automatically open the most-viewed video.

Start with videos that appear meaningfully above the channel's normal performance.

These are the videos most likely to contain new information about:

• Topic demand
• Packaging
• Timing
• Format
• Audience interest

Step 3: Check Whether the Signal Transfers

Look across other channels.

Ask:

• Did the topic work elsewhere?
• Are small channels winning with it?
• Is the winner recent?
• Are multiple independent creators confirming the demand?

An outlier becomes much more useful when the broader market also supports the hypothesis.

Step 4: Reverse-Engineer the Video

Once a competitor video survives the evidence checks, use OverseerOS Reverse Engineer to inspect the components separately.

Depending on the research question, that can include:

• Title
• Thumbnail
• Hook
• Transcript
• Outline
• SEO context
• Content structure

Do not copy the competitor.

Extract the mechanism.

Step 5: Turn the Evidence Into an Original Direction

The goal is not:

Their video got 3 million views. Make one like it.

The goal is:

This video was a 9x channel outlier, the underlying topic has cross-channel proof, the opportunity is still current, and the packaging reveals a mechanism we can adapt into something original.

That is a production decision you can defend.

The Metric Decision Framework

Use this before choosing a competitor video to study.

If Your Goal Is Finding Viral Video Ideas

Prioritize:

1. Outlier score
2. Cross-channel topic confirmation
3. Recency
4. Packaging quality

Do not prioritize raw views alone.

If Your Goal Is Finding Current Trends

Prioritize:

1. Recent velocity
2. Publication date
3. Views per day
4. Multiple current channels covering the topic

A large historical outlier may no longer represent a live opportunity.

If Your Goal Is Finding Ideas a Small Channel Can Win With

Prioritize:

1. Outlier score
2. Views per subscriber
3. Small-channel examples
4. Independent topic confirmation

The biggest channels in the niche may be the least useful benchmark.

If Your Goal Is Estimating Total Market Demand

Prioritize:

1. Raw views
2. Number of independent winners
3. Total channels covering the topic
4. Persistence over time

Here, absolute reach becomes much more relevant.

If Your Goal Is Studying Packaging

First verify that the video is an actual outlier.

Then inspect:

• Title
• Thumbnail
• Promise
• Curiosity mechanism
• Stakes
• Audience framing
• Hook alignment

A pretty thumbnail on a normal-performing video is not automatically a winning pattern.

We Repeated the Test With Different Rules

A result this strong deserves a sensitivity check.

Using Only the Previous 10 Videos as the Baseline

Instead of requiring 20 previous uploads, we repeated the study using a 10-video baseline.

That expanded the sample to:

• 3,710 videos
• 135 channels
• 15 niches

The central result remained.

Zero of 15 niches had the same #1 video under all four metrics.

Raw views and outlier score agreed on the #1 video in only:

2 of 15 niches.

Using a Tighter 30–180 Day Video Window

We also reduced the age window to focus on newer videos.

That produced:

• 2,277 videos
• 109 channels
• 14 niches

Again:

Zero of 14 niches had the same #1 video under all four metrics.

The exact rankings moved.

The central conclusion did not.

Metric choice materially changes which competitor videos appear worth studying.

What This Study Does Not Prove

This study does not establish that outlier score is universally "better" than every other metric.

That would misunderstand the result.

Outlier score is better suited to one specific question:

Which video most dramatically departed from what was normal for its own channel?

Raw views are better at measuring absolute reach.

Views/day is better at reducing age distortion.

Views/subscriber is useful for channel-size context.

The study also uses public data.

We do not have competitors' private:

• Impressions
• Click-through rate
• Audience retention
• Watch time
• Traffic-source breakdown
• Thumbnail experiments
• Returning-viewer data

Those private metrics could explain why a video performed.

Public competitor metrics can identify where to investigate.

They cannot recreate YouTube Studio for somebody else's channel.

Limitations

Several limitations matter.

First, this was not a random sample of every channel on YouTube. The data comes from public channels encountered through OverseerOS analysis and research workflows.

Second, we focused on long-form videos so Shorts and long-form did not share the same baseline.

Third, each primary video was between 30 and 365 days old. This reduces extreme age differences but does not give every video identical time to accumulate views.

Fourth, our views/day calculation is cumulative average views divided by age. It is not true recent views per hour.

Fifth, views/subscriber uses the channel subscriber snapshot available during analysis. A channel may have had a different subscriber count when an older video was originally published.

Sixth, our outlier calculation uses the median current public views of the previous 20 long-form uploads. That is a channel-relative research measure, not an official YouTube metric.

Finally, ranking metrics identify interesting videos.

They do not establish causality.

A 10x outlier may have succeeded because of:

• Topic
• Thumbnail
• Title
• Timing
• Audience
• External events
• Format
• Hook
• Distribution
• Several of those at once

The metric tells you where to look.

The research tells you what might transfer.

Final Verdict

Which YouTube competitor metric should you trust?

Trust the metric that matches the decision you are making.

In the OverseerOS analysis of 2,971 videos across 107 channels and 13 niches, all four metrics never selected the same #1 video in a single niche.

Raw views and outlier score selected the same #1 video only once.

Their top-10 lists overlapped by just 2.3 videos on average.

And they surfaced radically different types of channels:

• Raw-view winners came from channels with a median 2.18 million subscribers.
• Outlier-score winners came from channels with a median 122,000 subscribers.
• Raw-view winners were only 1.87x their channel baseline.
• Outlier-ranked winners were 9.91x baseline.

So if your goal is competitor research for your next video:

Do not ask which competitor video is biggest. Ask which video is unusually successful, still relevant, independently validated and transferable to your audience.

Start with the outlier.

Then add the context.

That is how competitor data becomes a content decision instead of a vanity-metric spreadsheet.

FAQ

What Is the Best Metric for YouTube Competitor Analysis?

There is no single best metric for every goal.

For discovering competitor videos worth reverse-engineering, channel-relative outlier performance is a strong starting point because it identifies videos that dramatically exceeded what was normal for that creator.

Raw views are better for absolute reach, views/day helps with age, and views/subscriber helps account for channel size.

Is Total View Count a Good Way to Find Viral YouTube Ideas?

It can identify large successful videos, but total views alone does not tell you whether that performance was unusual.

In the OverseerOS study, the median raw-view top pick came from a channel with 2.18 million subscribers and performed only 1.87x above its channel baseline.

A smaller competitor with fewer absolute views may contain a much stronger breakout signal.

What Is a YouTube Outlier Score?

A YouTube outlier score compares a video's performance with the channel's normal performance.

In this study:

Outlier score =
Video views
÷
Median views of previous 20 long-form uploads

A 5x score means the video received five times the median views of the preceding comparison group.

Is Views Per Subscriber Better Than Raw Views?

It is better for answering a different question.

Views per subscriber helps identify videos that reached far beyond the apparent size of their channel.

It does not necessarily tell you whether the video dramatically beat what that channel normally receives.

Is Views Per Day the Same as YouTube View Velocity?

No.

In this study, views/day means cumulative public views divided by the video's age in days.

True recent velocity measures how quickly views are being added during a recent time period.

The two metrics can tell different stories.

How Many Videos Should I Use to Calculate a YouTube Channel Baseline?

For the primary study, we used the previous 20 long-form uploads to establish the channel baseline.

We also repeated the analysis using 10 previous videos and found the central conclusion remained unchanged.

For a dedicated test of baseline stability, see our study on how many YouTube videos you should analyze.

Why Can a Video With Fewer Views Be More Valuable for Competitor Research?

Because the important question is often not absolute reach.

It is:

How surprising was the result relative to what this creator normally achieves?

A 300,000-view video from a channel averaging 20,000 may reveal a stronger new signal than a 3-million-view video from a channel averaging 2.5 million.

Should I Copy a YouTube Outlier?

No.

An outlier is a research lead.

Study:

• The topic
• Why the topic may have demand
• The title promise
• Thumbnail concept
• Hook
• Format
• Timing
• Cross-channel evidence

Then build an original version for your own audience.

What Should I Do After Finding a Competitor Outlier?

Validate it.

Check whether:

1. The topic has worked across other independent channels.
2. The result is recent enough to matter.
3. Smaller channels are also succeeding.
4. The packaging contains a transferable mechanism.
5. Your version can offer a genuinely different angle.

Then use the evidence to decide whether the idea deserves production.