Operator: Claude Code (Cowork, scheduled)
Window: Sunday 17 May 2026 (Day 0.5, last day before the formal Mon May 18 kickoff)
Mode: Local code writes only — push is manual (per task instructions). All edits live in the working tree at /Users/thunderbolt/Documents/projects/Shilika, ready to commit and push.
Scope: Ship the Run 2/3 staged title and meta rewrites (Day 1 task D1.4 pulled forward), bump dateModified across the site, lift /testimonials with a proper schema + intro, and refresh llms.txt to mirror the new title set.
Run 2 identified 6 over-length titles and 4 over-length meta descriptions; Run 3 confirmed the queue. Today I executed the entire copy-only sweep against the live repo. All 9 indexable pages now have titles ≤ 57 chars and meta descriptions ≤ 159 chars (under Google's ~60 / ~165 truncation thresholds). All 9 JSON-LD graphs revalidate. dateModified bumped to 2026-05-17 site-wide. /testimonials schema upgraded from a flat Person to a 4-entity graph (WebPage + BreadcrumbList + Person + WebSite) and the visible page got a 2-paragraph intro plus a tighter t-sub line. Sitemap lastmod bumped to 2026-05-17 on all 9 URLs. llms.txt headlines now match the new titles and a Last updated 2026-05-17 line was added.
Files touched (12): index.html, blog.html, testimonials.html, article-1.html, article-2.html, article-3.html, article-4.html, article-5.html, article-6.html, llms.txt, sitemap.xml (and the inline <style> block in testimonials.html).
Important context from research: Google published the AI Optimization Guide on Friday 15 May 2026 — two days before this run. Headline message: AEO and GEO are still SEO. The guide explicitly downplays special schema, llms.txt, and content chunking as growth levers. Foundation already shipped in Run 1 is correct and aligned; today's work intentionally stays surgical (titles, metas, dates, schema completeness) and avoids over-optimization patterns the guide warns against (no scaled long-tail page sprawl, no micro-paragraph chunking, no faux freshness).
All cut to ≤ 57 chars, keyword-first, no trailing brand suffix on long-tail pages (the brand suffix was the main length culprit).
| Page | Before (chars) | After (chars) |
|---|---|---|
| /article-1 | The Tier-1 trap: why most Web3 founders chase the wrong publications - Shilika Jain (83) |
The Tier-1 PR Trap: Why Web3 Founders Chase Wrong Outlets (57) |
| /article-2 | Korea, Japan, Vietnam: the APAC PR playbook nobody hands you - Shilika Jain (75) |
APAC PR Playbook 2026: Korea, Japan, Vietnam, India (51) |
| /article-3 | From technical founder to category voice: a 90-day profiling sprint - Shilika Jain (82) |
Founder Profiling Sprint: 90 Days to Category Voice (51) |
| /article-4 | How much does crypto PR cost in 2026? The honest breakdown - Shilika Jain (73) |
How Much Does Crypto PR Cost in 2026? Honest Pricing (52) |
| /article-5 | How to get featured in CoinDesk in 2026: a practitioner's guide - Shilika Jain (78) |
How to Get Featured in CoinDesk in 2026: Pitch Guide (52) |
| /article-6 | Fractional PR vs Web3 PR agency: how to choose in 2026 - Shilika Jain (69) |
Fractional PR vs Web3 PR Agency: How to Choose in 2026 (54) |
| /blog | Field Notes - PR essays for Web3 & AI founders - Shilika Jain (61) |
Web3 & AI PR Playbooks: Field Notes by Shilika Jain (51) |
Every title rewrite was mirrored across <title>, og:title, twitter:title, and (on blog.html) the headline field of each BlogPosting inside the Blog @graph.
All cut to ≤ 159 chars (under Google's ~165 truncation point), em dashes removed per user preference. All mirrored across og:description and twitter:description.
| Page | Before (chars) | After (chars) |
|---|---|---|
| / | 187 | Fractional PR for Web3 and AI founders. 50+ protocols placed in Forbes, CoinDesk, Cointelegraph, Decrypt and The Block across APAC. Book a 30-min teardown. (155) |
| /blog | 135 (was fine, refreshed for clarity) | Playbooks on Web3 and AI PR: Tier-1 outlet access, founder profiling, APAC strategy, crypto PR pricing, CoinDesk pitching. By a senior 2026 operator. (149) |
| /article-4 | 215 | What crypto PR actually costs in 2026: agency retainers, fractional models, single-launch sprints, KOL waves and the hidden line items most contracts bury. (155) |
| /article-5 | 201 | What CoinDesk editors filter for, pitch templates that land, embargo etiquette and three worked examples from a senior Web3 PR operator's 2026 playbook. (152) |
| /article-6 | 210 | Fractional PR vs Web3 PR agency in 2026: cost, speed, founder access, bandwidth and accountability. A decision tree that tells you which model fits your stage. (159) |
/testimonials upgradePerson + aggregateRating is now a 4-entity @graph: WebPage (with datePublished: 2025-04-01, dateModified: 2026-05-17, inLanguage, breadcrumb linkage) + BreadcrumbList + Person (carrying the 19-recommendation aggregateRating and 4 detailed Review entries) + WebSite. Closes the Run 2 audit gap (missing dateModified)..t-intro block under the existing kicker that names the buyer personas (founders of funded Web3 protocols, AI CMOs, exchange leads, Tier-1 journalists), the engagement length pattern (3 to 12 months), the verticals covered (DeFi, infra, agentic AI, NFTs, gaming, DePIN), and the 6 APAC markets. Matches the on-page-quality bar Google's May 15 guide emphasizes.UPDATED 2025 → UPDATED MAY 2026.