Where a YouTube video idea comes from, and when it is ready to film
Maxim Gavrilovich, founder of GoxiPublished 7 min read
On YouTube a video is decided before anyone picks up a camera. A viewer sees a title and a thumbnail and chooses in a second whether to click, so the idea has to work at that size first. Everything after it, the research, the script, the footage, exists to keep the promise the title and the thumbnail made.
When I first wrote the process down, it looked like this. Something happens that could be a video. From it you draft a title, a thumbnail and a short description. You improve them until you would click yourself. Then the idea goes into the shooting plan, and the script, the research and the B-roll come after.
That model mostly holds. It skips two steps, though, and blurs a third, and those gaps are where many weak videos start. This article walks through the whole path, from the first signal to the moment an idea earns its place in the plan.
What starts a video idea?
A reason to make it. Call it a signal: something you noticed that could become a video people want to watch. News is one kind of signal, and far from the only one.
- An event, a news story or a date that is coming up.
- A video on another channel that gets far more views than that channel usually does.
- A document, a public record or a story from an archive.
- A question your audience keeps asking in comments or emails.
- Something you did, tested or saw yourself.
A signal does not have to be fresh. A story from twenty years ago becomes a reason for a video the day you find new material about it, or the day people start asking about it again.
YouTube points creators to the same places. The Trends tab in YouTube Analytics shows what your audience and viewers across YouTube are searching for (YouTube Help). Starting from a signal, rather than from a blank page, is how the platform itself suggests you look for ideas.
Why is a signal not an idea yet?
Because one signal can carry several different videos. Between the signal and the thumbnail sits a step that is easy to skip: choosing the angle.
Say you come across the filings from the shutdown of a well-known startup.
| What you have | What it could become |
|---|---|
| Source material | Documents from the shutdown |
| Topic | How the company collapsed |
| First angle | Why its customers left |
| Second angle | Which warnings everyone ignored |
| Third angle | What happened to the money customers had paid |
Each angle needs its own research, its own script and its own title and thumbnail. Before you commit to one, check that the material supports it. The documents may tell a different story from the one you hoped to find.
An idea exists when it is clear who the video is for, what question it answers and why that viewer would want the answer. The title and the thumbnail make the idea concrete, and they show whether its promise can be understood at a glance.
What goes on an idea card?
Five things, all of them short:
- The signal and where it came from: a link, a document, a screenshot of the comment.
- The angle, in one sentence.
- A working title.
- A thumbnail sketch. A rough drawing is enough at this stage.
- A synopsis of a few sentences: what the video is about, where the conflict is, what the viewer will find out and what backs it up.
That synopsis is not the description under the published video, and the published description matters less at the idea stage. For the decision to click, YouTube's own advice centres on the title and the thumbnail (YouTube Help). The description helps viewers understand the video, helps it surface in search, and holds chapters, sources and links. Its first lines are the ones people see before they press "more", so write those for people, and fill in the rest once the video exists (YouTube Help).
How much research comes before the plan?
Split research in two.
The quick check happens before an idea goes into the plan. It answers three questions. Does the story hold up? Is there enough in it for a whole video? Can you show what the thumbnail promises? This should take hours, not days.
The deep research comes after the idea is chosen, together with writing the script and looking for footage. That is where you collect the details, the quotes and the scenes the video will stand on.
Mixing the two costs you either way. Do the deep research for every idea and you never film. Skip the quick check and you find out halfway through the edit that the story you promised is not in the material.
What does the whole process look like?
Like this, with one caveat: it loops back.
| Stage | What you have at the end |
|---|---|
| Collect signals | Links, observations, questions, documents |
| Quick check | The facts, the audience's interest, the material you can get |
| Shape the idea | A specific viewer, a central question and a promise |
| Package it | A few title and thumbnail pairs and a short synopsis |
| Pick for production | A decision that weighs interest, evidence, cost and time |
| Deep research and the script | The story, its evidence, the structure and a plan for the visuals |
| Film, edit, publish | A video that keeps the promise |
| Look at the results | What to change in the next package and the next idea |
The loops are normal. A document you find can change the central question. Missing footage can kill an idea. A strong scene from the shoot can suggest a better thumbnail. Plan the B-roll from the script stage on, because what you can show shapes how you can tell the story.
Is "I would click this" enough?
It is a good first filter. It is not proof that anyone else wants the video.
Early on, other people's successful videos are the best references you have. Read them carefully, though. How far does the video stand out from its own channel's usual results? How long did it take to collect its views? Is its audience the one you want? A million views on a channel that always gets a million tells you little. The same million on a channel that usually gets a tenth of that tells you a lot.
Following someone does not have to mean copying the video. Borrow the principle: the type of question, the way the investigation unfolds, the way something is explained. Then apply it to your own material.
After you publish, YouTube judges the whole chain, not the packaging alone. Its recommendation system looks at whether people chose to watch, whether they kept watching and whether they were satisfied: clicks, watch time, survey responses, shares, likes and dislikes all count, and the stated goal is long-term viewer satisfaction (YouTube Help, YouTube Blog). Even YouTube's own test of titles and thumbnails, which compares up to three versions on a published video, picks the winner by watch time rather than by clicks (YouTube Help). A package that wins the click and loses the viewer does not win.
When does an idea go into the shooting plan?
When you can say this about it, and mean it:
We know who this is for, we can express that interest in a title and a thumbnail, and we have already checked that we have the material for the video we are promising.
If any part of that sentence is shaky, the idea stays on the board for now. That is cheaper than finding out in the edit.
Where does Goxi fit in?
Goxi covers the first half of this path. It tracks the videos that beat their own channel's usual level among channels like yours, and it turns them into ideas for your channel: the topic, an opening hook, a title, a thumbnail concept and a short outline, with a drawn thumbnail draft for each idea. The quick check, the deep research, the shoot and the edit stay with you. How Goxi reads channels and judges what works is in How Goxi works.
What else do people ask?
Does every video need a news hook? No. A news story is one kind of signal. A question your viewers keep asking, a record you found or an experiment of your own can start a video just as well, and they age more slowly than news.
Should I copy a video that went viral? Copy the reason it worked, not the video. First check that it really stood out against its own channel's usual views. Then borrow the type of question or the structure, and apply it to your material.
Which comes first, the title or the thumbnail? They are one decision. Draft them together, a few pairs per idea, and keep the pair that makes the promise clearest. On a published video, YouTube can test up to three titles and thumbnails and keep the one viewers watch longest.
Does the description matter for clicks? Less than the title and the thumbnail. It helps viewers understand the video, helps search, and holds chapters, sources and links. Write its first lines for people, because they show before "more".
How much research should I do before I film? Enough to answer three questions: the story holds up, there is enough for a whole video, and you can show what the thumbnail promises. The deep research comes after the idea is chosen.