Competitive-landing-analysis skill — 100 competitors → hypothesis-ranked recommendations

A Claude Code skill takes a competitor list (~100 in your vertical), auto-assembles the required sub-skills (scraper, content extractor, comparison engine), and runs a multi-role prompt chain — CRO, UX designer, marketer, CMO — to produce a ranked HTML report of landing-page hypotheses. Optional weekly cron alerts you to changes on competitor pages.

Manual competitive landing analysis takes a week; agent-driven analysis runs in an afternoon and surfaces patterns a human reviewer misses across 100 pages. One marketing lead reports 3–4 quality recommendations adopted per analysis cycle. Re-runnable weekly, the workflow becomes a passive competitive-intelligence stream rather than a one-off project.

1. Define a competitor list of ~100 companies in your vertical (URLs of their primary landing or product page).
2. Ask Claude to assemble the required sub-skills (scrape page, extract structured content, summarise hero+CTA, compare).
3. Run the multi-role prompt chain — CRO analyzes conversion mechanics, UX designer analyzes layout/hierarchy, marketer analyzes positioning, CMO synthesises strategy.
4. Output a ranked HTML report with hypotheses scored by importance × implementation complexity.
5. Pick the top 3–4 hypotheses for the next landing iteration.
6. Optional: schedule a weekly cron run that diffs current state vs last week's snapshot and alerts on material changes (new hero copy, new pricing tier, new CTA).

A 100-URL run with naive scraping hits rate limits and gets blocked. Stagger requests; cache aggressively; rotate user agents. The skill is also only as good as the competitor list — adding 30 irrelevant competitors dilutes the synthesis. Curate the list before each cycle.