Guides
YouTube channel analytics: a monthly like-for-like routine
How to read YouTube channel analytics month to month: like-for-like comparisons, catalog versus new uploads, a format example and your own log.
What to take away
- A small swing between two months is not a finding. Studio's figures are estimates built on modeled data, and the platform says so.
- Treat differences between periods as noise until they persist.
- Compare like with like. A video published on the 28th has three days of data, while one published on the 2nd has a full month.
- Separate back-catalog watch time from new releases, because the catalog is the most stable number a channel has.
- Write down the change you think caused a movement before you check the data.
- Keep your own monthly record of views, watch time, RPM and revenue outside Studio, because Studio's retention windows and definitions can change.
This is general YouTube creator-business information, not platform, monetization, tax or legal advice. Eligibility rules change with your content, audience, location and channel status.
A small swing is not a finding
A 3% swing between two months usually means nothing. Studio's figures are estimates built on modeled data, and the platform says so. Treat small differences between periods as noise until they persist, and judge a movement only after it has repeated.
Pull the same four Studio views on the same day each month: Overview, Content, Audience and Revenue. Each holds a different kind of number, and mixing them is the most common mistake. Open all four before drawing a conclusion, and note the date you pulled them.
Four Studio views and their questions
View
- Overview
- What happened this period
- Content
- Which videos did it
- Audience
- Who watched
- Revenue
- What it paid
Question it answers
- Overview
- Content
- Audience
- Revenue
A view count with no watch time tells you a thumbnail worked and nothing else. Add watch time, then returning viewers, then revenue, and only then ask what moved. For how the views themselves read, see retention curves and traffic sources.
Compare like with like
Compare videos on equal footing or the ranking is wrong. A video published on the 28th has three days of data; one published on the 2nd has a full month. Sort by publish date before you rank anything, then apply four rules.
- Compare each video at the same age, not in the same calendar month.
- Split the catalog from new releases. Back-catalog watch time tells you whether old videos still earn.
- Read an outlier month twice. A viral short can dominate a month of watch time, so look with it and without it.
- Give a small channel a longer window, because returning viewers show noisy swings.
When a number moves, write down the change you think caused it before you check. That habit stops you from inventing a story after the fact. Then compare the month with the same month last year, not with a channel you do not run.
Where each of the five numbers misleads
Every Studio number answers one question and misleads on another. Click-through rate is clicks divided by impressions, so a low rate points at the thumbnail or title, not the video. Duration is time watched, and percentage viewed is that time divided by video length.
Five metrics and where they mislead
Metric
- Impressions CTR
- Did thumbnail earn click
- Average view duration
- Did video hold attention
- Watch time
- Total minutes watched
- Returning viewers
- Viewers who came back
- Revenue, RPM, CPM
- What it earned
What it answers
- Impressions CTR
- High CTR, nobody finishes
- Average view duration
- Long videos flatter raw number
- Watch time
- Compounds as catalog ages
- Returning viewers
- Rising views, falling returns
- Revenue, RPM, CPM
- Sponsorship income absent
Where it misleads
- Impressions CTR
- Average view duration
- Watch time
- Returning viewers
- Revenue, RPM, CPM
| Metric | What it answers | Where it misleads |
|---|---|---|
| Impressions CTR | Did the thumbnail earn the click | High CTR on a video nobody finishes |
| Average view duration | Did the video hold attention | Long videos flatter the raw number |
| Watch time | How much total attention the channel earned | A viral short can dominate a month |
| Returning viewers | Is the audience yours or rented | Small channels show noisy swings |
| RPM | Revenue per thousand views, all sources | Sponsorships and merch sit outside it |
A 20-minute video with 4 minutes watched and a 4-minute video with 3 minutes watched tell opposite stories. Read duration and percentage together and never alone, because each one hides what the other shows. A long video can look weak on percentage and strong on minutes, and a short one can show the reverse.
Worked example: two formats, one break-even
Format A runs 8 minutes and averages 3 minutes watched. Format B runs 22 minutes and averages 6 minutes watched. Format B earns more total watch time per video, and Format A earns a higher average percentage viewed, 37.5% against 27%.
Format A versus Format B
Format A
- Length
- 8 minutes
- Average watched
- 3 minutes
- Average percentage viewed
- 37.5%
- Total watch time per video
- Lower
Format B
- Length
- 22 minutes
- Average watched
- 6 minutes
- Average percentage viewed
- 27%
- Total watch time per video
- Higher
Whether to keep making Format B is not decided by which percentage is higher. It is decided by whether Format B's extra watch time covers its extra production hours. Write it as arithmetic with your own numbers, not with a borrowed benchmark:
- Format B extra watch time per video = B total watch time minus A total watch time.
- Format B extra production hours = B edit hours plus B shoot hours minus the same for A.
- Break-even = extra watch time divided by extra production hours.
If that figure is below what an hour of your time earns elsewhere, the format is a hobby with a good retention curve. For the money version of the same test, see the channel's break-even points. Which formats deserve a comparison at all is covered under formats to test.
A four-week routine
Build the habit in four weeks, one step at a time. Each week adds one check, and by the fourth you have a record you can compare against the same month last year without guessing at what changed.
- Week 1.Open all four Studio views and write down which numbers you already check and which you ignore. Note the last decision any of them changed.
- Week 2.Pick one number and one decision it should inform. Write the formula, the source and the review date on one page.
- Week 3.Build a simple monthly log outside Studio: views, watch time, RPM, revenue and returning viewers. One row per month.
- Week 4.Compare the month with the same month last year. Keep the metric if it changed a decision; drop it if it only added reporting work.
Before you trust a number
Run this list before you act on any movement. A change that survives all five checks is worth a decision; one that fails any of them is worth another month of data.
- The same four Studio views, pulled on the same day of the month.
- Videos compared at the same age, not in the same calendar month.
- Catalog and new releases reported on separate lines.
- Any outlier month read with and without the outlier.
- The suspected cause written down before the data was checked.
Keep the log outside Studio and decide which KPIs to track monthly. For the money lines, the IRS says you may choose any recordkeeping system suited to your business that clearly shows your income and expenses.
The log holds totals, not people. If it ever holds viewer-level data, the California Consumer Privacy Act is relevant, and what you may collect and retain is a question for a privacy lawyer.
Common questions
Why does my RPM change when my views do not?
RPM divides total revenue by total views, so it moves with ad rates, viewer geography, video length and how many views carried ads. A month with steady views and a falling RPM usually means the mix of viewers or the season changed, not that YouTube paid less per ad.
Should I compare my channel to benchmarks I find online?
Only if the benchmark names its source, its date and the channel type it measured. A figure without those three things cannot tell you whether your channel is underperforming or just different. Compare your channel to its own prior months first, then build your own baseline.
How far back should I keep my own numbers?
Keep them indefinitely, because Studio's retention windows and definitions change and your own log is the only record that stays comparable. A spreadsheet with one row per month costs minutes and settles arguments about whether a change worked.
Do sponsorships show up in YouTube Analytics?
No. Studio reports platform revenue only. Sponsorship, affiliate and merchandise income have to be logged separately, and the arithmetic that matters is total revenue across all sources divided by the hours the channel took to produce it.
