Robert W. Dyche IV

Credentials & Methodology Authorship

Robert's approach to GEO is grounded in first-principles measurement, not borrowed SEO playbooks. His methodology draws from the Princeton GEO paper (Aggarwal et al. 2023), Semrush AI traffic benchmarks, Microsoft Clarity behavioral patterns, and hundreds of raw matrices from active 2025-2026 programs. No post-hoc selection or vanity metrics — every number ties back to timestamped, re-tested prompts.

Day 0-90 Proof Cycle: The Signature Framework

Robert designed the four-phase cadence now used across all Stay Citable engagements. Each phase has a defined measurement gate:

1. Day 0 Baseline: Lock 30–50 high-intent prompts. Run across six AI answer surfaces. Record every citation, position, and source. No changes until the baseline is set.
2. Foundation (Days 0-30): Schema architecture (FAQPage, Article, Organization, Person/SameAs), content restructuring for AI readability, entity signal deployment. Early verification signals and Perplexity/Gemini traces typically appear here.
3. First Lifts (Days 30-60): 15-30%+ of target prompts surface brand citations by this checkpoint. Relative share-of-voice tracked against competitors on the same schedule. Adjustments made based on what's working.
4. Compounding (Days 60-90): Multi-engine presence established. At Day 90, we re-test the full prompt matrix and publish the results. Every number traces back to timestamped, re-tested prompts — no post-hoc selection. We publish what we achieve because the methodology is the proof.

Origin Story: How the Proof-Cycle Methodology Was Born

The Day 0-90 framework didn't start as a product — it started as a frustration. In early 2024, Robert was running traditional SEO programs for B2B clients and watching the same pattern repeat: ChatGPT and Perplexity were citing competitors that hadn't done any SEO work at all. The clients with the best backlink profiles, the most content, the highest DR — invisible. The ones getting cited had clean schema, clear answer structure, and entity signals the LLMs could parse.

He ran the first informal GEO experiment on a SaaS client in Q2 2024: lock 30 prompts, run them across ChatGPT and Perplexity, change nothing else for 30 days except schema + content structure, re-test. The result was 6 net-new citations in 30 days — a 20% lift with zero new content. That single experiment proved the hypothesis: AI engines reward structural clarity and entity signals differently than Google rewards backlinks. The methodology scaled from 30 prompts to 50, from 2 answer surfaces to 6, and from one ad-hoc test to the formal four-phase cadence used today.

By late 2024 the framework was producing consistent results in controlled experiments. The vacation rental vertical — tested via the cabin-rentals.us laboratory property — revealed a significant gap: most operators had zero AI presence, and the fix path (schema + llms.txt + answer-first content) showed strong early signals. The methodology was developed and tested before it ever appeared on a pricing page.

Research & Data Foundation

Robert personally runs the measurement for every engagement. Every claim published on this site traces to one of three sources:

• Controlled experiments: Timestamped prompt matrices from laboratory property tests, with full before/after documentation.
• Published academic research: Princeton GEO study (Aggarwal et al. 2023), plus follow-up research on AI citation behavior, schema impact, and cross-platform consistency.
• Industry benchmarks: Semrush AI traffic data, SparkToro zero-click data, G2 buyer behavior reports, and Microsoft Clarity behavioral analysis.

Why This Matters for E-E-A-T

When AI models decide who to cite, they evaluate the entity behind the content — not just the page it appears on. Google's Search Quality Evaluator Guidelines explicitly weigh creator expertise, authoritativeness, and trustworthiness. Robert's published methodology, timestamped results, and primary-source attribution give AI models and search engines exactly what they need to establish Stay Citable as a trusted entity in the Generative Engine Optimization space.