How to see which competitor reels beat their normal views and compare their hooks
Most competitor research ends with a folder of screenshots. A useful version ends with a table showing which openings worked, for whom, and how often. Here is how to build it.
Short answer. To compare competitors' reels, score each account's last 30 posts against that account's median views, keep the ones at twice the median or more, and get the transcript of each. Then put the first line, the first visual, the hook type and the score into one table. Sort by hook type. A hook type that appears across several accounts is a real pattern. Compare each competitor's outliers against its own normal posts as well, since that shows what actually made the difference.
Why look at outliers and not top posts
A competitor's most-viewed posts are mostly old and mostly lucky. You also cannot tell what made them work, because nothing about them is compared to anything. As one complaint about manual research put it, people end up "spending hours manually scrolling through competitors" and still have no clear result.
Scoring each post against its own account fixes this. An account with a median of 8,000 and a reel at 60,000 gives you a 7.5× outlier. That reel beat the same creator, same audience and same production. The difference is much more likely to be the idea or the opening. How to find outlier posts covers the maths.
Step 1: Choose four to eight competitors
Pick accounts that reach your audience, and include a few close to your size. A brand 100 times your size has a different job and different budget. If you serve a local or narrow market, include good accounts from other cities or close niches, as they will not mind and you can learn the same things.
Step 2: Score their posts
- Open the Reels tab and record views for the last 30 reels.
- Skip anything from the last week, since it is still growing.
- Find each account's median.
- Divide each post by its median.
- Keep anything at 2× or more.
Do this in a shared sheet with one tab per account. Expect three to six outliers per account, though very spiky accounts will have more and very steady ones fewer.
Step 3: Get the transcripts
You need the words as well as the views. There are a few ways to get them, in rough order of effort.
- On-screen captions. Many reels already show the spoken words, which you can type out for the first 10 seconds.
- Auto captions. Some apps show generated captions you can read while the video plays.
- Dictation. Play the reel and let your phone's dictation capture it, which is rough but quick.
- A transcript tool. XCut, Sandcastles and others take a link and return the transcript.
For a hook study, the first 10 seconds is enough. You only need the full transcript for posts you intend to study closely. Keep this for your own research, and do not republish someone else's words.
Step 4: Build the hook table
Here is the table to fill in. These rows are made up for illustration, for a fitness niche.
| Account | Score | First line | First visual | Hook type |
|---|---|---|---|---|
| A | 7.5× | "You are stretching wrong." | Creator mid-stretch | Direct accusation |
| B | 4.2× | "I did 100 squats a day for a month." | Date on screen | Personal experiment |
| C | 3.1× | "Your physio won't tell you this." | Close-up to camera | Hidden information |
| A | 2.6× | "Stop doing this after leg day." | Text over gym shot | Direct accusation |
Name each hook type in your own words and keep the list short. Ten or so types is plenty. Here are some common ones to start from.
- Direct accusation: tells the viewer they are doing something wrong.
- Personal experiment: "I tried X for 30 days."
- Hidden information: something insiders know.
- Result first: shows the end state before anything else.
- Specific audience: "If you are a nurse working nights..."
- Myth reversal: states a common belief and rejects it.
- Question: a question the viewer wants answered.
- Visual surprise: something odd or striking with little talking.
Step 5: Read the patterns
Sort the table by hook type and look at three things.
- Frequency across accounts. A type that shows up for four different competitors means more than a type used by one.
- Strength. Which types tend to produce the highest scores?
- Gaps. Which types does nobody in the niche use? A gap may be an opening for you, or it may be a sign that the type does not suit this audience. Test it before you trust it.
Step 6: Compare against their normal posts
This is the step that most research skips, and it gives you the sharpest insight. Pick the competitor's top outlier and three of its normal posts. If normal posts open with a greeting and the outlier opens with an accusation, that difference is your evidence. If they open the same way, the hook did not cause the result, so look elsewhere, such as at the topic or the format.
Without this comparison, you may credit the hook for what the topic did. A hook table alone can mislead you, since a good topic can carry an average opening.
Step 7: Look beyond the hook
| Also note | Why it matters |
|---|---|
| Topic | Often explains more than the opening |
| Length | Shows what this audience will sit through |
| Format | Talking head, list, demo or story |
| Text on screen | Many people watch on mute |
| Call to action | What they ask viewers to do |
| Top comments | What viewers took from it |
Turn it into something you can use
Finish with a one-page summary: your top three hook types, your top three topics, and five openings you could write for your own account. Do not reuse their lines. Write yours using your own subject and experience, as covered in how to turn a viral reel into your own script. Then test them against your own median.
Run a lighter version each month. Patterns shift, and an opening that was fresh last spring can feel tired once everyone is using it.
A hook table with real structure
This table is made up to show what a finished one looks like. The examples come from an imaginary personal finance niche.
| Account | Score | First line | Hook type | First visual |
|---|---|---|---|---|
| A | 6.2× | You are paying this bill twice | Hidden loss | Text over a bank app |
| A | 3.1× | Stop rounding up your savings | Direct order | Face to camera |
| B | 8.4× | I checked 40 of my old receipts | Experiment | Pile of paper |
| C | 2.7× | Nobody told me this about rent | Secret | Face to camera |
| C | 4.9× | The cheapest month of my life | Story | Empty fridge |
Add a row for each outlier. After 20 or so, sort by hook type and see which types beat their median most often. Then look at the first visual, because it is easy to miss. In this made-up table the two highest scores open on an object or a screen and not a face. A pattern like that might point to something you would never see if you only read the transcripts.
Ten hook types to label with
Use one simple set of labels so your table stays comparable. This is a working list that we find practical, not a standard. Add your own as you go.
| Type | What it does | Made-up example |
|---|---|---|
| Mistake | Warns about an error | Stop washing your rice like this |
| Hidden loss | Shows what they are losing | Your phone plan is costing you double |
| Result | Opens with an outcome | This took my page from 2 to 200 leads |
| Myth | Challenges a belief | Stretching will not fix your back |
| Question | Asks what they already wonder | Why does your dog pull on the lead? |
| Experiment | Sets up a test | I tried every budget app for a month |
| Story | Drops into a moment | The day my landlord called |
| Secret | Promises inside knowledge | What estate agents will not tell you |
| Order | Tells them what to do | Delete this app today |
| Identity | Calls out a group | If you are a new manager, watch this |
What the transcript will not show
Viewers decide using more than the words. One creator noted, after feedback: "looks like have to improve more on the first 3s hook in the reel". Those first seconds include several parts, so review each hook on four layers.
- Spoken: the first sentence.
- Text on screen: often a different message from the spoken line, and it is what people read with the sound off.
- Visual: what is happening in the frame, and whether it moves.
- Sound: a voice straight away, or music, or a noise that grabs attention.
Watch each outlier twice with the sound off and once with your eyes closed. It takes a few minutes, and it shows you which layer is carrying the hook. Then note whether the rest of the video delivers what the opening promised. A hook that overpromises may get people in, but they leave quickly, and in our view that works against you.
How to avoid getting stuck on competitors
There is a cost to this research. One creator admitted: "Yeah, I think I've been looking at competitors way too much." Another wrote: "I've been always struggling and always checking out competitors". Comparing can become a way of feeling behind instead of learning something. These limits keep it useful.
- Set a date. Once a month is plenty for most creators.
- Set a time limit, such as 60 minutes.
- Leave with three things to test and not a page of notes.
- Study the work and not the follower count.
- Look outside your niche as well. Hooks from another field will look fresh to your audience.
Ad libraries are worth adding, too. As one commenter suggested: "TikTok Creative Center + Meta Ads Library swipe file." Brands pay for those ads, so ads that run for a long time may be working, although you cannot see their results. Treat long-running ads as a hint and not as proof.
Test the hook before you film the whole video
Filming is the expensive part. Write five openings for your next post, using different hook types. Read them to someone who knows nothing about the topic, and ask which one they would keep watching. If you want a stronger test, film the same video with two different openings and post them a few weeks apart. It will not be a clean experiment, but over time you will build a record of which hook types work for your own audience. The free hook generator can give you options to react to when you are stuck.
Doing this at scale
For four accounts, a sheet is fine. For agencies or anyone tracking many accounts, the manual work quickly becomes the whole job. XCut AI scores posts against each account's median across Instagram, TikTok, YouTube and Facebook, and reads the video to lay out hook, angle, structure, pacing and call to action. Because the sources sit side by side on a canvas, you can compare them directly, then draft from them. The free competitor gap finder is a quick first look. For other options, see the best tools to find viral reels.
Frequently asked questions
How can I see a competitor's reel views?
Public play counts appear on the Reels tab of most profiles, though a creator can hide some counts. You cannot see another account's private insights such as saves or reach.
Is it OK to get transcripts of competitors' reels?
Studying public posts for research is common practice. Do not republish their words as your own. Check the platform terms if you plan to collect data at scale.
How many hooks do I need to see a pattern?
Around 15 to 20 outlier hooks across at least four accounts is a reasonable start. Fewer than that and one creator's style can look like a pattern.
Should I copy a competitor's best hook?
Copy the type and not the line. Write a new opening about your own subject, then test it against your own median.
How often should I repeat this?
A light pass monthly works for most accounts. Add to the same table so you can see which hook types are rising or fading.
Should the text on screen say the same thing as the spoken hook?
Not always. Some creators use the text to add a second reason to stay. Check what your competitors' outliers do, and test both ways.
How long should a hook be?
There is no fixed rule. Keep it as short as it can be while still saying who the post is for and what they will get. Your own retention data will tell you if people leave early.
References
- Outlier detection: what makes content go viral.
- Best Instagram competitor analysis tools.
- Free XCut AI tools.
See which of your posts beat your average
Paste your handle at xcut.ai/instagram for a free scan with no account, or start a trial to research any niche.
Start your 7-day trial$5 for 7 days · 1,000 credits · cancel anytime