A competitor teardown studies repeatable choices—format, promise, proof, and audience response—without taking someone else's expression. It helps you see what a market already explains well and, more importantly, what questions remain unanswered.

By the end, you will have a three-account teardown with five pattern takeaways, a gap map, and one original experiment. It is research, not a shortcut to another creator's identity.

// concept

Choose a Comparable Three-Account Sample

Choose three public accounts serving a similar audience that are one stage ahead in production consistency—for example, publishing on schedule or using a format you are learning. This does not reveal revenue or business health.

Fix the sample before drawing conclusions:

Use one platform and a fixed period, such as the previous four complete weeks.
Include every relevant post, not only posts you liked.
Mark five posts with the strongest visible response within each account's sample.
Mark two weaker-within-sample posts for comparison.
Manually log public URLs, dates, formats, topics, hooks, promises, proof, and response. Never access private dashboards, closed groups, leaked reports, or personal data, or automate collection against platform rules.

Public likes, comments, or views do not reveal sales, audience quality, paid distribution, total reach, or cause. Never call a post a business “winner” or “flop” from counters alone.

Keep one sheet row per post and one tab per account.

Field	What to record
Account and post	Account label, public URL, publication date
Sample position	Top-five public response, weaker-within-sample, or other
Format	Talking head, carousel, screen demo, interview, list, story
Promise	The specific outcome or understanding offered
Proof	Demonstration, source, process view, customer evidence with consent, or none
Response	Visible signals plus date observed; summarize comment themes separately
Gap	A relevant question asked but not answered in the post or recent feed

// concept

Separate Patterns From Protected Expression

Compare rows rather than assets. A pattern is abstract: “screen demo after a problem-first hook” or “five-slide checklist with a caveat on slide two.” Distinctive expression includes exact wording, sequences, jokes, illustrations, thumbnails, or footage.

Patterns and gaps yes — scripts, jokes, and thumbnails no. Credit is not permission. To show another creator's post, obtain permission or qualified advice; the safer workflow records the pattern privately and builds fresh material.

For each account, summarize five lenses:

Format: Which structures recur, and which appear in stronger versus weaker rows?
Promise: Does the opening offer a result, explanation, comparison, warning, or story?
Proof: Does the post show the process or cite evidence, or only assert a claim?
Response: What questions or objections appear in public comments? Record themes, not usernames or copied comments.
Gap: Which repeated audience question is not answered by the sampled posts?

Write five cross-account takeaways as observations, not rules: “In this four-week sample, screen demonstrations appeared more often among stronger-within-account rows.” Add a counterexample where possible.

// concept

Turn a Gap Into a Distinct Experiment

A useful gap joins an audience question with evidence you genuinely have. Ask: Can I answer it accurately? Do I have permission for customer material? Can I create a different promise and proof method?

After redacting names, paste only your notes into this prompt:

// prompt — copy me17 lines

You are helping me analyse a manual sample of public content.
Do not infer revenue, sales, reach, intent, or algorithmic causes.
Do not reproduce any source wording, joke, script, or visual composition.

Audience I serve: [precise segment]
My evidence/experience: [what I can demonstrate or source]
Sample period and platform: [dates and platform]
Observed patterns: [your abstract pattern notes]
Unanswered public-comment themes: [paraphrased and anonymised]

Return:
1. five observations supported only by these notes;
2. gaps that I can answer with my stated evidence;
3. three original experiment ideas, each with a different format,
   promise, proof method, and response signal;
4. a copying-risk check for each idea.
Label uncertainty. Do not claim likely business results.

Make one experiment distinct on at least three dimensions. For example, change a talking-head warning into a screen-recorded walkthrough, replace “avoid mistakes” with “compare two options,” and use your own demonstration as proof. Predefine one content signal, such as qualified questions or completion rate in your own analytics. It informs the next test; it does not prove revenue.

// worked_example

Worked Example

The following is a hypothetical sample, not a report about real accounts or performance. A Lahore-based accessories seller studies three unnamed public Instagram accounts for the previous four complete weeks. The audience is first-time online buyers who struggle to judge product size from photos.

They fill the analysis prompt with this actual input:

// prompt — copy me4 lines

Audience: Lahore first-time online accessories buyers.
Evidence I own: current product dimensions and an original measurement demo.
Observed gap: public comments ask how to measure at home.
Generate distinct experiments; do not infer sales or reuse source expression.

The sheet shows these sample observations:

Account	Repeated format	Promise	Proof	Public response theme	Gap
A	Talking-head tips	Avoid ordering mistakes	Product held beside common object	Questions ask for exact measurements	No measurement walkthrough
B	Product carousel	Compare available sizes	Labelled photos	Questions ask how to measure at home	No home method shown
C	Packing video	See what arrives	Unedited packing sequence	Questions ask about fit before ordering	No decision guide

An AI draft proposes: “Copy Account A's opening line and recreate its product-in-hand thumbnail, then promise the perfect size every time.” This takes distinctive expression, makes an absolute promise, and ignores the gap.

The seller fixes it with an original experiment: a screen-and-hands tutorial titled “Measure your current pouch before choosing a new one.” It uses the seller's own ruler, product, footage, and measurement chart. The hook is a question—“Will your charger fit?”—the proof is a live three-step measurement, and the caption states that the viewer must compare the result with the current listing dimensions. The planned signal is the number of relevant sizing questions in the seller's own comments; no sales outcome is predicted.

The resulting gap-map entry reads: “Question: how to measure at home. Existing coverage in sample: unanswered. Evidence we own: current dimensions plus original demonstration. Distinct treatment: measurement tutorial, not a talking-head warning or product carousel. Rights check: original script, footage, objects, and thumbnail.”

// failure_cases

Failure Cases to Diagnose

6 cases to diagnose

Cherry-picked sample
You logged only popular-looking posts. Fix it by defining dates first and including stronger and weaker rows from every account.
Cross-account counter comparison
You rank a large account against a small one using raw likes. Compare posts within each account and describe public signals without declaring performance causes.
Business inference
“This format drives sales” appears without private conversion data. Replace it with the exact observation your sheet supports.
Surface-level gap
“They have not made a blue carousel” is not an audience need. Look for repeated unanswered questions or missing evidence.
Disguised copying
Synonyms were swapped into the same hook, joke, script sequence, or thumbnail. Return to the underlying problem and create new wording, structure, examples, and visuals.
Private-data collection
The sheet contains DM screenshots, customer phone numbers, or closed-group posts. Delete them; use public posts and anonymised public-comment themes only.

// pakistan_angle

Pakistan Angle

Pakistani audiences vary by city, region, language, device, and buying habit. A Karachi English-Urdu fashion sample should not be presented as evidence for Pashto-first buyers in Khyber Pakhtunkhwa. Record the language mix and region the account appears to address, then test your own angle with the audience you actually serve. Avoid reducing “Pakistani relatability” to chai, rishta, or traffic jokes.

Many small brands sell through Instagram and WhatsApp, where comments may contain phone numbers, delivery addresses, or order details. Do not copy those into a teardown or AI prompt. Redact CNICs, numbers, handles, and faces; never treat a public comment as permission to feature a customer. If load-shedding or mobile data interrupts research, save post URLs and text notes first, then review media later on Wi-Fi rather than downloading and redistributing competitors' files.

// hands_on

Hands-On Exercise

7 steps

Define one platform, one precise audience, and the previous four complete weeks as your sample.
Select three comparable public accounts and write why each is one stage ahead.
Log all relevant posts, then mark five stronger and two weaker public-response rows per account.
Complete format, promise, proof, response, and gap fields without copying source expression.
Write five cross-account pattern takeaways, each tied to rows and qualified as sample observations.
Build a gap map with the audience question, existing coverage, evidence you own, distinct treatment, and rights check.
Draft one original experiment with a new hook, structure, proof asset, and a signal you can observe on your own account. You are done when one document contains the three account tabs, five supported takeaways, the gap map, and the distinct experiment brief.

// completion_rubric

Completion Rubric

6 checks — tick as you verify

0/6

// sources

Sources

4 official sources — check every claim yourself

// check_yourself

Check yourself

4 questions · answers and options are taken word-for-word from this course

0/4

1 / 4 · diagnose
Your work shows this failure mode: “Cherry-picked sample.” The lesson describes it like this: “You logged only popular-looking posts.” What does the lesson tell you to do about it?