Most video briefs spend pages on brand tone and half a line on the opening. That is backwards. The opening is where the viewer decides to stay or scroll, and it is the part creators most often get wrong.
You can fix that without guessing. Every niche has videos that beat their account's usual results by a wide margin, and their openings are public. Collect them in one place, tag them, and you have a hook library built from evidence.
A hook library is a tagged collection of the opening lines and opening visuals from videos that beat their own account's median views. You build it by finding outliers, writing down the first line word for word, tagging the hook type and the promise, and counting what repeats. Then you write your own openings from the patterns, not the wording.
Why build it from outliers
A hook is only as good as the way people respond to it, and the response you most want to see, who stayed past the first seconds, is private to the account owner. What you can see is the outcome. A video that far beats its own account's median probably held attention, and its opening is a good candidate for the reason. We explain that reasoning in our research on outliers and retention.
Starting from outliers keeps the library honest. You are not collecting hooks you happen to like. You are collecting hooks attached to results. Use the outlier score to decide what goes in, and a 2× cut-off is a sensible place to start.
Seven hook types worth tagging
These are craft categories, not statistics. They give you a vocabulary for spotting patterns. The example lines are illustrative, written to show the shape, and are not taken from real videos.
| Hook type | What it does | Illustrative opening line |
|---|---|---|
| 1. Result first | States the outcome before explaining anything | "I cut my editing time from six hours to forty minutes." |
| 2. Contrarian claim | Challenges something the viewer believes | "Posting every day is the reason your reach is dropping." |
| 3. Curiosity gap | Hints at a finding without giving it away | "The one setting that quietly cut my views in half." |
| 4. Direct question | Asks something the viewer wants answered | "Why do your best videos get fewer views than your worst?" |
| 5. Specific call-out | Names the exact viewer it is for | "If you sell candles online, stop doing this." |
| 6. Comparison | Sets two options against each other | "The $12 version against the $120 version." |
| 7. Number list | Promises a bounded, countable payoff | "Three mistakes I see in every first draft." |
Illustrative lines, to show the shape of each type.
Real videos often combine two types, such as a specific call-out wrapped around a result. That is fine. Tag the main one, and note the second in the comments column.
How to build the library
1. Collect outliers, not favourites
Pick videos that scored 2× their account's median or more. Aim for 20 to 30 across at least five accounts in the same niche. Fewer than that and one account will dominate the patterns.
2. Watch the first three seconds, then write it down exactly
Record the opening line word for word, and describe what is on screen in the first second. The exact words matter, because the patterns hide in them.
3. Tag the hook type and the promise
Choose one of the seven types. Then write the promise in one sentence: what the viewer is told they will get.
4. Log the context
Add the video's length, platform and outlier score, so you can see whether a hook type is tied to a length or a platform.
5. Count what repeats
After 20 to 30 entries, count the hook types and the promises. Patterns that repeat across accounts are the useful ones. A hook that appears once is a curiosity.
6. Review it every month
Hooks go stale as more people use them. Add new outliers monthly and retire patterns that stop showing up.
The spreadsheet
One row per video. These columns are enough:
How to read it
- Count by hook type. If four of ten outliers in your niche open with the result first, that is your starting hypothesis.
- Pair type with promise. The combination matters more than either on its own. A comparison hook about price may work where a comparison about speed does not.
- Look at what is missing. If nobody in the niche uses a direct question, either it does not work there or nobody has tried it. Both are worth knowing.
- Check length. Some hook types need time to pay off, and some work best in a short video.
How to use it in a brief
For each idea, write five openings from the patterns, not copies of the wording. Choose two to test. Keep the body of the video the same and change only the opening, so that you learn what the opening did. Score each result against your own median, the same way you scored the references. Our guide to copying a viral video explains how to separate what to reuse from what to leave alone, and our 30-day UGC research system shows how hooks fit into a full brief.
How Statly fills the library for you
The slow part is finding the outliers. Trend Finder shows the videos beating their account's median across every niche you track on Instagram, TikTok and YouTube, refreshed daily, so you are not scrolling to find them. Save the ones you want to study to your Library, open an account in Watchlist to see whether the same opening appears again and again, and export any table to CSV to start your sheet.
Questions and answers
What is a hook library?
A hook library is a collection of opening lines and opening visuals taken from videos that beat their own account's median views, tagged by hook type and promise. You use the patterns, not the exact wording, to write your own openings.
How many hooks should be in a hook library?
Start with 20 to 30 outlier videos from at least five accounts in the same niche. Fewer than that and one account will dominate the patterns. Add new outliers every month and retire patterns that stop appearing.
What are the main types of video hooks?
Seven useful categories are result first, contrarian claim, curiosity gap, direct question, specific call-out, comparison and number list. They are craft categories for spotting patterns, not statistics, and many videos combine two.
Can I see a competitor's hook rate?
No. Hook rate and retention are private to the account owner. What you can see publicly is whether a video beat its own account's median, which is the closest outside signal that the opening held attention.
Should I copy the exact words of a winning hook?
No. Use the hook type and the promise as patterns and write your own words. Copying exact wording is less original, can run into copyright and platform rules, and does not tell you what actually worked.
How do I test a hook?
Keep the body of the video the same, change only the opening, and post two or three versions. Divide each video's views by your own median and compare the scores. Judge on several tests, not one.