Competitor teardown

Competitor Teardown

Produces a one-product teardown across four lenses — positioning, pricing, onboarding, gaps — where every factual claim carries a source URL fetched during this run. The standard: observed beats reported beats inferred, each claim is labeled which of the three it is, and the deliverable ends with what the user should do about it, not just what the competitor does.

When to Use

• User names one competitor (or 2–3, run sequentially) and wants to understand how they win.
• Before a positioning, pricing, or onboarding decision: "what does X do here?"
• Investor/accelerator prep: "who else does this and how are you different?"
• Not for: market size math (market-sizing-tam); recurring signal monitoring (trend-radar); auditing a page's SEO (seo-content-audit).

Quick Reference

Action	Command / Call
Homepage positioning	web_extract on the root domain — capture hero headline, subhead, social proof, CTA verb
Pricing	web_extract on /pricing — plans, prices, billing basis, gated features, free tier
Onboarding (observed)	web_extract signup page + docs "getting started"; note steps, required fields, time-to-value claims
Onboarding (reported)	search "<competitor>" onboarding OR "getting started" review — G2, Reddit, YouTube walkthroughs
Sentiment & gaps	search "<competitor>" site:reddit.com and "<competitor>" alternatives — recurring complaints
Recent moves	search "<competitor>" changelog OR "what's new" — last 90 days

Procedure

1. Precheck — confirm web tools are available. Get from the user: the competitor name + URL, the user's own one-line positioning (needed for the gap analysis), and which lens matters most right now. No user positioning? Ask — a teardown without a "compared to what" produces trivia.
2. Positioning — extract the homepage. Record verbatim: hero headline, subhead, the customer named ("for teams who..."), proof elements (logos, counts, testimonials), and primary CTA verb. Then write the implied positioning in one sentence: for <who>, X is the <category> that <differentiator>. Label it inferred — it's your read, not their words.
3. Pricing — extract /pricing. Capture every plan: name, price, billing basis (monthly vs annual-billed — they differ), unit, free tier/trial, and the 3–5 features gating each step up. "Contact sales" is data — record as custom. Note the plan the page visually pushes; that's their target customer telling on itself.
4. Onboarding — extract the signup page and getting-started docs: steps to first value, required fields, credit-card-required?, self-serve vs sales-gated. Supplement with reviews and walkthroughs for the parts behind the signup wall — label those reported, with source.
5. Gaps — search reviews, Reddit, and "<competitor> alternatives" pages for recurring complaints (≥2 independent sources before it counts). Cross-reference with what the user does well. A gap is only exploitable if customers complain about it AND the user can credibly serve it.
6. Deliver — fill the template. Every claim row cites its URL and access date, and carries observed / reported / inferred. End with three moves for the user, each tied to a specific finding.

Output Template

# Teardown — <competitor> — <date>
Their positioning (inferred): For <who>, <X> is the <category> that <differentiator>.

## Positioning        | evidence: <url> (<date>)
- Headline (verbatim): "<...>" | Named customer: <...> | Proof: <...>

## Pricing            | evidence: <url> (<date>)
| Plan | Price | Billing | Key gates |
|---|---|---|---|
Pushed plan: <which and why that matters>

## Onboarding
- Steps to first value: <n> — <list> [observed: <url>]
- Friction / walls: <...> [reported: <url>]

## Exploitable gaps
1. <gap> — complained about at <url1>, <url2> — user can win here because <...> (confidence: high/med/low)
2. ...

## Your three moves
1. <specific action> — because <finding #>

Pitfalls

• Marketing copy read as ground truth — "loved by 10,000 teams" is a claim, not a fact. Recovery: label homepage numbers reported (their claim); only independently verifiable facts (prices on the page, steps you observed) get observed.
• Pricing page JS-blocked or extraction returns nav chrome — no numbers come back. Recovery: try the Wayback Machine (web.archive.org/web/2026/<url>) or review-site pricing data; mark all such numbers reported, unverified <date> and never present them as current.
• Gap analysis without the user's own position — you list competitor weaknesses the user can't exploit either. Recovery: step 1 requires the user's positioning; every gap must pass "customers complain AND user can serve it".
• One angry Reddit thread becomes 'users hate X' — a single source inflated into a pattern. Recovery: gaps need ≥2 independent sources; one source gets reported as "single mention, unconfirmed".
• Stale teardown reused — pricing and positioning change monthly. Recovery: date every source; if asked to reuse a teardown older than 60 days, re-fetch pricing and homepage before answering.

Verification

• Every factual claim has a source URL fetched this run, with access date
• Every claim labeled observed / reported / inferred
• Pricing table states billing basis for every price
• Each gap has ≥2 independent sources and a confidence label
• "Your three moves" each trace to a specific numbered finding