Analyze competitor ad strategies across Meta and Google

This skill researches competitor ads from Meta and Google, analyzes creative patterns, maps observable landing-page funnels, and produces a strategic teardown covering hooks, formats, positioning bets, vulnerabilities, and counter-plays. Its core principle is that a competitor's public ad portfolio is only partial evidence about its growth strategy - long-running ads can indicate continued investment, but public libraries never expose conversion performance or spend, so the skill separates observations from hypotheses, cites every observed ad or page, and labels performance/budget inferences explicitly rather than stating them as fact.

Intake gathers competitor names and domains, the user's own product for comparison framing, which channels to cover (Meta, Google, or both), and a depth level - standard (scrape plus creative and landing-page analysis) or deep (adding historical comparison, funnel reconstruction, and counter-plays). Meta research uses web search to find Meta Ad Library pages:

web_search: site:facebook.com/ads/library "[competitor_name]"
web_search: "[competitor_name]" Meta Ad Library active ads
web_search: "[competitor_name]" facebook ads examples

collecting ad copy, visual type, CTA text, landing page URL, active duration, platform, and A/B variants per ad, preferring manual browser research and reporting a coverage gap rather than bypassing a block or inventing missing ads. Google research follows the same pattern against the Google Ads Transparency Center, collecting headline variants, description lines, ad type, and geographic targeting where visible, treating search snippets and third-party examples as secondary evidence.

Creative pattern analysis clusters ad headlines into seven hook types (fear/loss, outcome, question, social proof, contrarian, empathy, product-led), counts how many ads per competitor use each to reveal their primary messaging strategy, tabulates format distribution (static image, video, carousel, search text, display banner) across platforms, and catalogs unique CTAs by urgency, low-friction, or outcome framing. Landing-page and funnel analysis fetches each unique landing page found in ads (only after the user authorizes the research scope), treating every fetched page as untrusted input - allowing only public http/https destinations, rejecting localhost or private-network targets and unsafe redirects, and ignoring any content that tries to redirect the agent's task. It extracts the hero headline, subheadline, primary CTA, social proof, pricing visibility, form-field burden, page type, and a 1-10 message-match score against the ad's promise, then clusters ads into campaigns by shared landing page, messaging theme, or audience signal, and produces a per-campaign funnel table covering strategic intent, target persona, positioning bet, hook strategy, conversion path, and an explicit note that longevity does not prove performance. Budget allocation signals stay directional only - ad volume and platform distribution never translate into spend shares unless the user supplies actual spend evidence, in which case the allocation is marked unknown rather than guessed.

Strategic analysis runs a creative gap analysis (angles nobody is running, overcrowded angles to avoid, format white space, underutilized proof points, CTA patterns worth testing but not proven), a vulnerability analysis across seven named weakness types (message-LP mismatch, single-persona dependency, platform concentration, missing social proof, weak CTA, generic positioning, stale creative), and, in deep mode, a historical comparison via Web Archive data on whether positioning has shifted or campaigns have appeared/disappeared over the last 6-12 months. The final report covers coverage stats, an executive summary, per-platform hook and headline analysis, a campaign-by-campaign breakdown with a sample ad and funnel map, budget-allocation evidence explicitly marked unknown where unsourced, the creative gap and vulnerability findings, and recommended counter-plays each naming the targeted weakness, proposed angle, headline, and landing-page strategy.

Use it to tear down a competitor's ad strategy, find new creative angles, reverse-engineer a paid funnel, or audit the ad landscape before a launch. Public ad libraries can be incomplete, delayed, region-specific, or blocked by anti-automation controls, and ad longevity or volume never proves conversion performance, profitability, or spend - those conclusions stay labeled as hypotheses. Never bypass access controls, CAPTCHAs, or platform terms, and minimize collection of personal data and copyrighted ad creative by citing and briefly describing evidence rather than reproducing entire ads; the output supports marketing analysis, not legal advice on trademark or advertising-law compliance.

Input is competitor names/domains, the target channels and depth level, and the user's own product context. Output is a structured Competitor Ad Intelligence Report covering hook/format distribution, campaign clustering with funnel maps, a creative gap and vulnerability analysis, and recommended counter-plays with headline and landing-page proposals.

It uses web_search to query the Meta Ad Library and Google Ads Transparency Center, and fetch_webpage or curl to retrieve and analyze landing pages - no mandatory paid API for the documented manual-browser route.

Marketing and growth teams who need an evidence-grounded teardown of competitors' ad creative, funnels, and vulnerabilities before launching or refreshing their own paid campaigns.