Real-Time Intelligence + Distribution — Growth Estimates & Authority Moat

This layer makes ScamCheck real-time and self-reinforcing: trending detection feeds the public pages, distribution pushes alerts out, embeddable widgets pull links back, and the authority graph compounds topical strength — without leaving the free/hobby cost envelope.

What shipped

• Live trending engine (trending() + viral/momentum/active detection) surfaced on every scam page via a static-first TrendingStrip that hydrates from a CDN-cached snapshot API (one cheap read behind the cache).
• Real-time distribution (formatAlert): deterministic X/LinkedIn/WhatsApp/Telegram/Shorts snippets — no AI cost — via /api/scam-intel/alert-formats.
• Authority graph + hubs: seasonal (festival, tax), event (election), and entity (UPI, job/investment) hubs that interlink type pages and concentrate link equity; in-season detection for freshness boosts.
• Freshness layer: viral/active badges, last-reported timestamps, computeTrend decay, 72h "active alert" window.
• Embeddable backlink widgets (/widget.js, /widgets): one-line embed injected into the host DOM → real dofollow "Powered by ScamCheck" backlink; auto-updating; near-zero cost.
• Growth analytics (/api/scam-intel/growth): Discover candidates, fastest-growing entities, topic-leverage ranking.

Architecture & cost (static-first)

• Public pages stay SSG; real-time data arrives via client fetch of CDN-cached APIs → no rebuilds, negligible Firestore reads (snapshot is materialized by the daily cron).
• Distribution + widgets are deterministic / cached → no Vertex usage on these paths.
• Everything fits Vercel Hobby + Firestore Spark.

Analytics signals (for prioritisation)

• Highest-CTR topics: surfaced via Discover-candidate scoring (viral + active + severity + recency).
• Highest-RPM topics: investment, loan-app, tech-support (advertiser demand) — flagged in topicLeverage.
• Discover candidates: fresh + viral + high-severity clusters (discoverCandidates()).
• Fastest-growing entities: platforms/regions/categories by recent share (fastestGrowingEntities()).

Backlink potential

Widgets convert ScamCheck from a destination into distributed infrastructure other sites embed:

• Target embedders: regional news, consumer-finance blogs, bank/fintech help centres, RWAs/community sites, cybersecurity newsletters.
• Each embed = a persistent dofollow link; widgets update themselves, so embedders keep them.
• Planning estimate: even modest adoption (50–300 embedding domains in 6–12 months) yields dozens to a few hundred referring domains — the single biggest off-page lever for a new authority site. The /widgets page itself targets the "free scam widget/badge" query space.

Discover growth potential

Discover rewards freshness + entity authority + image quality. This layer adds all three: live trending/active badges (freshness), the hub/entity graph (authority), and (from the prior phase) dynamic OG images. Combined, expect trending scam pages to become Discover-eligible within weeks and Discover to grow into a material, spiky traffic source (Discover traffic is bursty by nature). Seasonal hubs capture predictable spikes (festival/tax/election).

Authority growth (the moat)

The moat compounds from four reinforcing loops:

1. Coverage — programmatic pages across types × cities × banks × platforms.
2. Freshness — real-time trending keeps pages "alive" and re-crawled.
3. Links — embeddable widgets pull in referring domains.
4. Citations — official-source attribution (1930/cybercrime.gov.in/RBI/SEBI) earns E-E-A-T and AI-Overview citations. Each loop feeds the others; competitors copying one loop don't get the compounding.

Highest-leverage traffic opportunities (do first)

1. Tier-1 type pages × Discover (UPI/OTP/KYC/phishing) — demand + freshness + AI-Overview structure already in place.
2. Widgets distribution — pitch the "Latest scams in India" widget to regional news/finance sites; highest backlink ROI.
3. Seasonal hubs ahead of the season — publish/promote festival & tax hubs 3–4 weeks before peak.
4. High-RPM topic pages (investment/loan/tech-support) — monetisation-weighted.
5. Hindi variants — large underserved query space, near-zero marginal cost.

All estimates are planning figures, not guarantees — outcomes depend on indexing, competition, embed adoption, and Discover's per-user ranking. Track actuals via /api/scam-intel/growth and Search Console.