$8.86M in attributed organic revenue, +217% traffic, and an SEO function that ended up running itself.
A 4-year embedded engagement at a regulated health brand. AI-assisted content production at scale, AI-search infrastructure built from scratch, and by year four, technical SEO running autonomously through a multi-agent workforce.
- Inherited flat organic at ~57K visits/mo and thousands of thin "YouTube-companion" URLs dragging the domain.
- Built and ran an AI-assisted content system on 7,000+ video transcripts. 2,074 articles published through editorial + FDA/FTC compliance review.
- Shipped foundational AI-search infrastructure (llms.txt, A2A, NLWeb, citation tracking) + off-platform LLM authority across Reddit, Quora, Wikidata, HARO, and Medium.
- By year four, the technical SEO function was running autonomously through a multi-agent workforce against WordPress VIP via GitHub and the REST API.
- Result: $8.86M attributed organic revenue ($6.02M in clean GA4 window), +217% traffic, DR68 → DR77, hundreds of thousands of monthly AI citations, zero compliance incidents.
A regulated health brand with massive video authority but flat organic.
Dr. Berg Nutritionals leads keto and intermittent fasting education with 15M+ YouTube subscribers and 7,000+ educational videos. When I joined in May 2022, organic search was flat at ~57K monthly visits despite the brand presence.
The core issue: thousands of indexed “YouTube-companion” URLs — pages whose only content was a video embed and a short description. They accumulated link equity, ranked for nothing, converted nothing, and dragged the domain's topical authority down.
Brief: build organic into a defensible acquisition channel without breaking FDA/FTC compliance discipline.
Six workstreams, sequenced deliberately.
Diagnose and sequence the work
- Built the “battleground clusters” framework: where Dr. Berg had real authority + where competitors were vulnerable.
- Sequencing: topic authority first, entity trust in parallel, buyer-intent capture last.
- Held the team to the order across multiple budget cycles — the part most teams skip.
Build the AI-assisted content engine
- Ingested Dr. Berg's 7,000+ video transcripts into a structured content production system.
- Every article: editorial review → FDA/FTC compliance QA → schema + on-page → publish.
- AI as scale layer on top of the editorial process, not as a replacement for it.
Kill the wrong approach mid-flight
- Initial wave was product-focused. Compliance flagged FDA/FTC issues on health claims.
- Killed product-focused content as a direction. Pivoted the whole operation to topical-authority.
- Same channel, different approach. The pivot saved the program.
Ship the AI search infrastructure
- Deployed llms.txt, A2A endpoints, NLWeb, and AI-friendly robots.txt.
- Implemented MedicalWebPage schema, author/reviewer markup, lastReviewed fields.
- Built citation share-of-voice tracking covering the major AI search platforms.
Off-platform LLM authority building
- Targeted authority placements on the platforms LLMs cite from.
- Wikidata entity completed programmatically by AI agents.
- Detail in the section below.
Automate the technical SEO function
- By year four, the technical SEO function was fully autonomous.
- Agents pushed changes to WordPress VIP through GitHub + WordPress REST API.
- Detail in the “Year 4” section below.
2,074 articles. Editorial + compliance gate on every one.
- Topic selection driven by AI citation gaps and authority cluster maps — not flat keyword volume.
- Citation magnets: symptom maps, root-cause frameworks, evidence tables, diagnostic tools built to be cited by humans and AI alike.
- Cross-cluster architecture: canonical authority pages, FAQ clusters, mechanism explainers, symptom pages, decision pages.
- Continuous re-optimization of top-traffic content fed back by performance data.
Built to be retrieved, not just ranked.
- Foundational AI-discoverability: llms.txt, A2A endpoints, NLWeb, AI-friendly robots.txt — so AI crawlers can ingest the site cleanly.
- MedicalWebPage schema + author/reviewer markup + lastReviewed fields — credentialed authorship signals for both Google and LLMs.
- Citation share-of-voice tracking across the major LLM platforms: share of mention, recommendation rate, citation quality, entity framing, competitor substitution.
- Top-25 prompts audited monthly. Full-100 quarterly.
Showed up where the LLMs actually look.
AI search platforms don't just crawl your domain. They weight authority signals from the open web — communities, Q&A sites, knowledge bases, news outlets. We built presence on every one that LLMs are known to cite from.
A diversified, compliance-safe program.
- Competitive link analysis: cross-referenced the top competitors' link profiles. Targeted the sites all of them had in common.
- PR campaigns every few months: each producing 100–300+ contextual placements across indexed news and trade outlets.
- Foundational links + profile building: branded niche profiles, Web 2.0 platforms, image surfaces, social signals.
- Tier 1 + Tier 2: a layered link velocity model that looks natural to Google's algorithm and the AI ingestion pipelines.
- Continuous profile health + proactive disavow management via GSC.
By year four, the SEO function was running itself.
For the final 6–12 months of the engagement, the entire technical SEO function was operated by a multi-agent workforce. The agents had write access to the production environment.
What was automated
- Keyword research end-to-end, including intent mapping and cluster assignment.
- Technical audits: crawl health, indexability, broken links, redirect chains, schema validation.
- On-page SEO: titles, metadata, internal linking, schema deployment.
- Content updates + reformatting pushed live via REST API and GitHub commits.
- Reporting + monitoring running on a continuous loop.
- GA4 audit fixes: the 90-item audit findings remediated automatically.
A working prototype of the end-to-end autonomous flow on the keyword “benefits of potassium.” Keyword research, structured SEO brief, content drafted to Dr. Berg's voice and editorial standards, interactive Gutenberg blocks, infographic auto-generated via Puppeteer, calculator + quiz, and structured publish — every element produced by AI agents and written code.
Four years. Clean numbers.
Headline outcomes
From the people who worked alongside me.
A highly skilled SEO professional who played an important role in strengthening our organic presence. Consistently demonstrated a strong understanding of content strategy, search intent, and how to build scalable SEO initiatives.
One of the most well-rounded SEO professionals I've worked with — a deep understanding of what drives sustainable organic growth, with the strategic thinking to back it up.
What's generalizable, what was situational.
Generalizable
- Kill-or-fix framing on underperforming sub-channels — fixing the approach often beats killing the channel
- Topic authority before commercial intent on YMYL brands
- AI-search infrastructure (llms.txt, A2A, NLWeb, citation tracking) applies to any brand
- Off-platform LLM authority via Reddit, Quora, Wikidata, HARO is the new link-building
- Compliance as routing decision, not blocker
Situational
- Dr. Berg had unusual upstream asset density (7,000+ videos) — most brands don't
- 4-year engagement is rare — most engagements get 12–18 months
- The autonomous-agent layer required a year of editorial baseline to train against
Let's talk about your organic channel.
If you're extending SEO into AI search, or deciding whether to fix or kill an underperforming acquisition channel, that's the conversation I want to have.
SEO Strategist & Agentic AI Specialist · Founder, Holistic SEO Company & Agentic Workforce Company
10+ years in SEO · 18 across digital marketing and operations · Virginia Beach, VA
More engagements
