Instagram competitor analysis with Claude: outliers, hooks, Meta ads and comments

Updated

An Instagram competitor analysis compares a rival's Reels and TikToks with their own median views to find their outliers, then breaks down the hooks, formats, Meta ads and comments behind them. With Claude connected to Virlo, 7 prompts turn a competitor's handles into a swipe file and your next 10 videos. Run on The Pink Stuff, it found the same burnt pan clip did 30.3M views on TikTok but 130K on Instagram, below its 165K Instagram median.

  • The 7 parts of the analysis and the swipe file tab each one fills
  • What it found on The Pink Stuff: hands-only demos, and one clip that hit on TikTok and died on Instagram
  • All 7 prompts, word for word
TikTokInstagram

What does an Instagram competitor analysis cover?

A complete Instagram competitor analysis covers seven things: the competitor's recent videos, their outliers, the hooks and formats behind those, their Meta ads, the people in their comments, a saved swipe file and a content plan for you. Each step below is one Claude prompt.

The 7 steps and what each produces
StepWhat it findsSwipe file tab
1. Pull their videos90 days of TikToks and Reels with views, comments, shares, savesVideos
2. Flag the outliersVideos at 3x their own median or more, plus repeat seriesOutliers
3. Hooks and formatsSpoken, on-screen and visual hooks, format, pacingHooks and formats
4. Meta Ad LibraryLongest-running ads, angles, offers, landing pagesMeta ads
5. CommentsBuyers, doubts, requests, complaints, hooks to stealComments, Creators
6. Swipe fileOne .xlsx with every tabAll
7. StrategySteal, skip and gap lists, plus your next 10 videosNext videos
Claude running a competitor analysis on The Pink Stuff step by step: TikTok and Instagram videos, outliers, hooks, Meta ads, comments, creators and the swipe file

How do you set up Claude for Instagram competitor analysis?

Claude needs live Instagram and TikTok profiles, comments and the Meta Ad Library, which the Virlo connector provides alongside 23.3M viral videos. Setup takes about 10 minutes.

  1. Open an account with the 7-day free trial.
  2. Add the connector: in claude.ai open Customize, Connectors, + Add, Add custom connector. Name it Virlo, paste the URL below, click Add and sign in. See Virlo for Claude for screenshots.
  3. Turn on code execution (Settings, Capabilities, "Code execution and file creation") so Claude can build the .xlsx file, then switch Virlo on in a new chat.
Connector URL
https://dev.virlo.ai/api/mcp/mcp

Then name your competitor. The original resource packs the 7 prompts into one Claude skill that runs from this request; without it, paste the prompts below in order.

Start the analysis
Run a competitor analysis on [BRAND].
TikTok: [HANDLE]. Instagram: [HANDLE]. Ads country: [US].
My accounts: TikTok [HANDLE], Instagram [HANDLE].

What the connector reads on the Instagram side is listed on the Instagram MCP page.

What did an Instagram competitor analysis find on The Pink Stuff?

Run on The Pink Stuff, the cleaning brand, the analysis set its usual video at 91K views on TikTok and 165K on Instagram, then found what its biggest outliers share.

The Pink Stuff's biggest outlier videos: burnt pan marks 30.3M on TikTok, cosy season clean 14.5M on TikTok, filthy finds 4.6M on Instagram, back to school 3.8M on TikTok
  • Hands only. No face and no voiceover on any of the 4 biggest videos. The product is already on the dirt in frame 0.
  • Same upload, different app, different result. The burnt pan clip did 30.3M on TikTok and 130K on Instagram. Judge a format per app.
  • The comments hand you the angle. The top comment on the 30.3M video (528 likes): "My pink stuff doesn't work like this .." A rival's first video writes itself: a side-by-side test.
The Pink Stuff competitor card: what to steal, skip and fill
ListItemWhy
StealHands-only demo, product on the dirt at frame 0Proven across their biggest videos; needs a phone and a sink, not their fame
StealA named weekly series ("Weekly Wins: Filthy Finds")Hit on both apps
StealA seasonal job with a plain promise ("The cosy season clean you can't skip")Re-check after 30 days; this one took a month to catch
SkipTheir song choices"this f*cking song" is the top comment on the 14.5M video
Gap"My pink stuff doesn't work like this .."Answer it with a side-by-side test
Gap"Sadly it's way too expensive for me to buy"Answer it with cost per clean
Gap"omg what is that first radiator one called?"Name the product on screen

The first video to make from it: a 12-second, hands-only test of the burnt pan clip with the product name on screen, opening on the burnt pan with your product already on it and paying off by second 2.

How do you find a competitor's outlier Reels?

Pull 90 days of the competitor's videos on both apps, then flag anything at 3 times that account's own median. Comparing to their median, never their follower count, is what makes an outlier real. Patterns with 3 or more outliers are Proven; 1 or 2 are Early.

1. Pull their videos
Using the Virlo connector, run lookup_creator twice for [BRAND]:
- platform tiktok, username [TIKTOK HANDLE]
- platform instagram, username [INSTAGRAM HANDLE]
Set include to "videos,outliers", max_videos to 100,
outlier_threshold to 3 and data_intelligence to true.

The hook, format and visual fields land a few minutes after the
lookup, so re-read each lookup before you use those fields.

Keep only videos from the last 90 days. If a platform has under
30 videos left, page through raw_creator_videos for that handle
until you reach 90 days back.

Leave out any video marked as paid. Give me one table per platform:
date, link, length, views, comments, shares, saves, first line
of the caption. Never invent a number.
2. Flag the outliers
From those tables, work out each platform's own median views.
Flag every video at 3 times that median or higher as an outlier.
Compare each video only to this account's own median, never to
its follower count.

Then give me:
1. The outliers, biggest first: link, views, multiple of their median
2. The series or repeat formats they post (same opening words,
   same structure), with the median multiple of each series
3. Any video posted on both TikTok and Instagram, side by side,
   and which ones hit on one app and died on the other
4. Length of the outliers against the rest
5. Three pairs from the same series or format: one outlier and one
   under their median. For each pair, name the one thing that's
   different (the job shown, the first frame, the text hook, the
   topic, the moment it was posted).

Label every pattern Proven (3 or more outliers) or Early (1 or 2).
One big video is a story, so it can't be Proven on its own.

To scan many smaller rivals for outliers instead of one brand in depth, use competitor content analysis.

How do you break down a competitor's hooks and formats?

Read each outlier's spoken, on-screen and visual hook from the data fields (never the caption), then label format, delivery and pacing. The test that matters most: would it still work without their fame, budget or mascot?

3. Break down the hooks and formats
For every outlier, use the hook, content format and visual
fields from the lookup. Take the hook from the hook field, never
the caption. For TikToks, also run raw_transcript so we know
what's said in the first 3 seconds.

For each outlier give me:
- Spoken hook, on-screen text hook and visual hook (frame 0).
  Check them in that order, since the spoken line does most of
  the work when there is one
- Hook type: question, how-to promise, teaser, big claim,
  relatable moment or talking straight to the viewer
- Format, in the words a marketer would use (talking head,
  before and after, hands-only demo, UGC testimonial, skit, listicle)
- On screen: face presenting, hands or POV only, or no person
- Product: featured, incidental or absent
- Delivery: conversational, high energy or instructional
- Pacing: length and rough number of cuts
- Does the hook's promise start paying off in the first 2 seconds?
- Does the end of the video close the hook's promise?
- Goal: reach (short, one idea) or saves and comments (longer,
  a list or a how-to)
- Would it still work without their fame, budget or mascot?
  Answer yes or no, and say why
- Why it beat their median, in one plain sentence, naming the
  hook type or topic. Mark anything you can't prove as a guess.

End with the 3 hook patterns and 2 formats they repeat most
across their outliers.

To see how a competitor's hook types compare with the wider market, check types of hooks.

How do you check a competitor's Meta ads?

Find the competitor's verified page in the Meta Ad Library and sort their active ads by how long they've run. Nobody keeps paying for a losing ad, so 90+ days live is as near to proof as the Ad Library gets.

4. Check the Meta Ad Library
Run raw_advertiser_search for "[BRAND]", country [US], and pick
the verified page that matches their Instagram handle. Then run
raw_advertiser_ads with that page_id, status active, country [US].
Page through up to 5 pages.

Give me:
1. Ads live longest, oldest start date first. Mark 90+ days as
   Proven, since nobody keeps paying for a loser. Collapse
   near-duplicate ads into one concept before you count anything
2. The main angles and offers, grouped, with how many ads use each
3. Where the ads send people: landing pages, sorted by ad count
4. Any ad that reuses a hook or video from their organic outliers
5. Angles they run in ads but never post organically, and the reverse
6. How serious this advertiser is: active ads and how many creatives
   sit behind one concept. A page with a handful of ads is testing.
   Hundreds of active ads means the winners are worth studying

The Ad Library doesn't show spend or views on most ads, so never
estimate them.

More on reading the library by hand: how to use the Meta Ad Library, or browse ads inside Virlo's Meta Ads Library.

How do you find buyers in a competitor's Instagram comments?

Sort every comment on their biggest outliers into five buckets. People asking where to buy, or asking for a product, are your next customers; doubts and complaints are the objections their videos never answer.

5. List the people commenting
Run raw_comments on their 5 biggest TikTok outliers and their
5 biggest Instagram outliers, 2 pages each.

Sort every comment into one bucket:
- Wants to buy (where to get it, price, does it ship to...)
- Doubt ("doesn't work", "mine didn't do this")
- Asks for a product or feature
- Complaint (price, smell, the audio, quality)
- Praise

Give me:
1. Bucket totals and the 5 most-liked comments in each, word for word
2. A list of people in "Wants to buy" and "Asks for a product":
   username, comment, video link. These are your next customers.
3. The 3 objections their videos never answer
4. The 10 comments worth stealing word for word as hooks, since
   nobody writes a hook better than the customer already did

Then run find_videos with topics [BRAND], creator_type individual,
last 90 days, sorted by most_viral, 2 pages. List the creators
posting about them: handle, platform, video link, views, what the
video is (review, ranking, test, tutorial). These are your next
creator partners.

How do you turn the analysis into a swipe file and content plan?

Save everything to one spreadsheet, then run the same lookup on your own accounts and compare. The output is three lists (steal, skip, gap) and your next 10 videos, each tied to a proven outlier.

6. Save it all into a swipe file
Build an .xlsx swipe file called "[BRAND] competitor swipe file"
with these tabs:
Videos | Outliers | Hooks and formats | Meta ads | Comments |
Creators | Next videos

Rules: one row per item, links in their own column, views as
plain numbers, freeze the header row, and sort Outliers by
multiple of median. Write only what came back from the tools.
If a tab has no data, leave a one-line note on it saying why.

Give me the file to download so I can open it in Google Sheets
(File, Import, Upload).
7. Turn it into a content strategy
Now run the same lookup on my accounts: TikTok [MY HANDLE],
Instagram [MY HANDLE]. Work out my median and my outliers the
same way.

Compare me to [BRAND]:
1. Hooks and formats they win with that I never use
2. Things I do better (keep doing these)
3. Objections and requests from their comments that nobody is
   answering. These are the openings.

Sort everything into 3 lists:
- STEAL: Proven patterns that passed the "works without their
  fame" test. Copy the format and hold everything else constant.
- SKIP: anything that only worked because of fame, a giveaway,
  a celebrity or a news moment you can't repeat.
- GAP: openings from the comments and ads they leave empty.

Then write my next 10 videos, each with: the outlier it comes
from (link), on-screen text hook, spoken hook, frame 0, format,
length, and one line on why it should work for my audience.
Change the topic and the proof to mine, so it reads as our own
video. Rank the 10 by how close they are to a proven outlier.

Then run it on your top 3 competitors, one chat each, and ask Claude to compare the 3 sheets. If all 3 land on the same format, your niche has told you what to film next. To audit your own account the same way, use the social media audit template, and track Reels over time with Instagram analytics.

Rather skip Claude? Run the same competitor research inside the Virlo web app.

How this resource was made

Source
Virlo, read 9 November 2026: The Pink Stuff's tracked TikTok and Instagram posts (34 per platform for the medians), live views, and the first 50 comments on its top 5 TikToks.
How this page is made
Written by the Virlo team from the original Competitor Analysis System resource, its 7 prompts and its Pink Stuff worked example.
Limitations
The Pink Stuff run kept to free reads, so it shows no Meta ads results; prompt 4 is the step that pulls them. Views are as of 9 November 2026.
As of
(first published November 9, 2026)

Source: Virlo, Instagram competitor analysis with Claude: outliers, hooks, Meta ads and comments, November 2026, virlo.ai/resources/instagram-competitor-analysis

Want the full list?

Turn a competitor's TikTok and Instagram handles into their outliers, hooks, Meta ads and buyers, saved to a swipe file. Virlo is social listening for TikTok and Instagram Reels: start a 7-day free trial at app.virlo.ai and see real videos from your niche before you pay.

Frequently asked questions

How do you do an Instagram competitor analysis?

Pull 90 days of the competitor's Reels, flag those at 3x their own median views, break down the hooks and formats, check their Meta ads and comments, then turn it into steal, skip and gap lists for your own content.

What is an outlier in a competitor analysis?

A video at 3 times or more the competitor's own median views on that platform. Comparing to their median, not follower count, shows which ideas truly worked.

Should I analyse Instagram and TikTok separately?

Yes, judge each app on its own median. The Pink Stuff's burnt pan clip did 30.3M on TikTok but 130K on Instagram, below its 165K Instagram median.

Can Claude analyse a competitor's Instagram?

Yes, with a connector such as Virlo that reads public profiles, Reels, comments and the Meta Ad Library.

How do I see a competitor's Meta ads?

Search their verified page in the Meta Ad Library and sort active ads by start date. Ads live 90+ days are the closest thing to proof, since nobody keeps paying for a loser.

What can competitor comments tell me?

Who wants to buy, what people doubt, what they ask for and which objections the brand never answers, plus customer phrases worth using as hooks.

What do I get at the end?

An .xlsx swipe file with tabs for videos, outliers, hooks and formats, Meta ads, comments, creators and your next 10 videos.

How many competitors should I analyse?

Start with your top 3, one chat each, then ask Claude to compare the sheets and look for a shared format.

How long does the setup take?

About 10 minutes: connect Claude to Virlo, turn on code execution so it can build the spreadsheet, then name the competitor's handles.

Run this on your own niche

Turn a competitor's TikTok and Instagram handles into their outliers, hooks, Meta ads and buyers, saved to a swipe file.

7-day free trial, easy to cancel.