How to audit your own channel

An audit sounds like something done to you by somebody else, with a spreadsheet and a serious expression. Most of it is one afternoon, your own analytics page and a notebook, and the hard part is not getting at the numbers. It is deciding what you are looking for before you open anything.

Look for strengths, faults and unanswered questions. A useful audit connects each observation with a possible explanation and a practical next step.

Why audit at all?

To find the pattern you cannot see from inside it.

You made every video on that channel, so you remember each as a separate afternoon. The day the tripod fell over. The one you filmed twice because of the bin lorry. Nobody watching experienced any of that. They experienced a body of work with shapes in it, and noticing those shapes means treating your videos as a set rather than as a run of Saturdays. That is all an audit is, and it is why you can run one on yourself.

How do you score a channel without scoring the videos?

By topic first, then sub-topic, and not by individual video.

Individual videos are noisy. One went out the week everybody was away, one got shared by somebody with a large following, one was uploaded at a strange hour because you fell asleep. Topics average that out. Write down the four or five subjects your channel covers, in the words you would use to a stranger rather than the words in your titles, then split each into the two or three sub-topics underneath.

Start with comparable work on your own channel, allowing for age, topic, length and traffic source. Relevant external benchmarks can add context, but should not replace those differences.

Mark each topic and sub-topic out of ten. The number is not the output. Being made to commit to one forces you to decide whether something is working or whether you enjoyed making it, and those two feelings are harder to tell apart than anybody admits.

Which of your videos already worked, and why?

This is the step people skip, and it earns the afternoon.

For every topic that scored well, write a sentence explaining it. Not that it did well. Why. Was it the most basic question in your subject, asked plainly? Was it the one where you showed the thing going wrong before you showed it going right? Was it the only video that answered a question somebody had put to you?

A strong result is an observation, even when the reason is unclear. Write a possible explanation as a hypothesis and identify what further evidence would support or weaken it.

Where do people leave?

Open the retention graph on your three strongest videos and your three weakest, and read the shape rather than the number at the end. You want to know whether the line drops in the same place across several videos. A fall in the first seconds usually points at what you promised on the way in; a slow slide through the middle usually points at pacing. Treat those as the first place to look in the footage rather than the verdict, because no one shape maps cleanly onto a single cause.

A repeated shape is worth investigating. An unusual single video can also reveal a question; neither establishes the cause on its own.

Which thumbnails were clicked?

Sort your videos by click-through rate, which is the platform's way of telling you how often somebody shown the thumbnail decided to open it, and read your own list from the top, comparing like with like: click-through moves with where a video was shown, so a search hit and a video pushed onto the home feed sit on different scales, and the read that holds is within one traffic source rather than across the whole list.

Do not hold the rate against any published figure. Read the list against itself. Put the top five images side by side, say what they have in common, then do the same with the bottom five. The answer is usually dull: a face doing something, one legible object, an unusual colour. It came from your audience rather than from a category, which is why it beats any rule anybody could hand you.

New and returning viewers

Your analytics separate people who had watched you before from people who had not, and they tell you which videos gained subscribers. Those two lists rarely match.

A video that pulls in strangers is doing the finding. A video your existing audience turns up for is doing the keeping. Different jobs, both legitimate, and a channel that only does one stalls in a way the view count never explains. Mark each topic as one or the other before you close the laptop.

High scores need explaining too

Strengths and faults answer different useful questions.

An underperforming subject may need a better package, clearer explanation, different distribution or less attention in the plan. A strong result may suggest something to repeat. Write the possible action and the evidence still needed for each.

The explanations on the high scores take longer to write than the ones on the low scores, and they are the part that changes what gets filmed next. It is also the limit of auditing yourself: you can score the pattern and explain your own wins, but you cannot read your channel from outside it, against what the rest of the category is doing.

What comes out of the afternoon

One page. Four or five topics, their sub-topics, a mark each, a sentence each, and a note saying whether each topic finds people or keeps them.

Then the decision the audit was for. Of the topics that scored well, which one could you make ten more videos about without repeating yourself or getting bored? That is your next quarter, and everything else on the page is context for that single choice.

Sources and evidence

What here is a platform fact and what is our own method. Platform behaviour was checked on 4 September 2026.

Standards and platform facts. YouTube's help pages define click-through rate as how often a viewer watched after seeing a thumbnail, and describe retention and the wider performance data as evidence of viewer response rather than proof of why any change occurred. That supports reading a shape across several videos rather than treating one number, or one graph, as a verdict. The pages define the metrics; they do not tell you how to read your own channel with them.

Our method. Scoring by topic rather than by video, comparing only against your own average, and writing a sentence explaining every good score as well as every bad one, is this studio's method rather than a platform instruction.

References.

• YouTube. Understand your YouTube content performance. support.google.com. Performance data describes viewer response; it does not by itself establish why a change occurred. Official explanation remains observational.
• YouTube. Understand your content performance. support.google.com. Retention and reach metrics have defined meanings. Definitions do not establish causal explanations.