Sign In

Ethan Smith™

Growth Advisor

Ethan Smith is the CEO of Graphite, a boutique growth agency specializing in building scalable growth engines for consumer tech companies.

With a strong background in advising and board membership, Ethan has worked with renowned companies like MasterClass, Honey, Ticketmaster, and Thumbtack.

Previously, Ethan held advisory roles at MasterClass, Mucker Capital, Ticketmaster, and Thumbtack.

He also served as the CEO of Sachi and as the Chief Growth Officer at Yummly before its acquisition by Whirlpool.

Moreover, Ethan has experience as a Board Member at BayAreaSearch and as the Director of Product at Yummly.

Ethan Smith brings a wealth of expertise in growth strategies and business development to the tech industry, evident through his diverse roles and affiliations.

Highlights

Jun 27 · linkedin

"AI Answer Optimization" & SEO

  • AI Answer Optimization is the new SEO
  • Reforge Presentation: https://lnkd.in/d9kKvJSy
  • Slides: bit.ly/ai-optimization
  • Graphite is launching AI Answer Tracking - https://lnkd.in/d9w8XQjh

Key Points

  • AI Search is increasingly used for search and discovery
  • Graphite is building strategies to help track and optimize for AI Search
  • Optimization for AI Search will likely be similar and an evolution of traditional SEO vs. something entirely different

LLM vs. LLM+RAG

  • Early versions of LLMs were based on next-word prediction, which is different from search
  • More and more, AI chat and search use RAG+LLM
  • RAG (retrieval augmented generation) starts with a search (retrieval), then summarizes/reforms the search as an answer (generation)
  • Perplexity has built its own search engine, CoPilot frequently uses Bing, and OpenAI is reportedly building a search engine - so AI Search is moving towards RAG+LLM more and more
  • Therefore, optimizing for AI Search starts with optimizing for search using traditional SEO strategies + is an evolution of new optimization strategies

Keyword Research > Question Research

  • "Question research" is the new keyword research
  • Rather than finding keywords, instead we need to find all the ways people ask questions for our product
  • We may target thousands of keywords, but questions vary more than keywords, thus there are millions of variations of questions
  • Keyword research can use known data from Google and other keyword tools
  • There are no tools to find what questions people ask - new strategies will need to be made to source how people ask questions
  • Based on the questions people ask, we create landing pages and content to target questions similar to SEO

Good Content for Questions

  • "Good content" in Google SEO is comprehensive and answers the questions users have for that page via the presence of TF-IDF terms, sub-topics, and embeddings
  • A landing page that targets a topic in AI Search needs to first understand the thousands of questions users have for that topic, then answer as many of them as possible
  • Example - ResortPass's West Hollywood Edition page should answer questions about reserving a cabana, how much does a day pass cost, and what are opening hours

Citation Optimization

  • RAG+LLM performs a search, looks at multiple pages, summarizes them, and cites its sources
  • Companies can optimize for being cited using SEO strategies similar
  • Forbes and NerdWallet can try to optimize for being a citation for "best credit cards" across many variations of credit cards

SERP Tracking > AI Answer Tracking

  • SEO tracks single positions for keywords (e.g. I rank #5 for "best credit card")
  • AI Answer Tracking is a distribution or frequency across surfaces, question variants, and question runs
  • There are no tools that allow companies to track their presence in AI Answers
  • Graphite is launching AI Answer Tracking - https://lnkd.in/d9w8XQjh
Jun 18 · linkedin

AI Content & Search - Study

  • José Luis Paredes, Gregory Druck, and I conducted a study @ Graphite to assess the prevalence and effectiveness of AI content in Google Search
  • We found that 88% of results in Google have little to no AI content, only 12% have high amounts of AI content, and only 3% are fully AI content
  • Further, we found that AI content typically underperforms human-written content
  • This contradicts the perspective that Google is becoming flooded with AI content and that mass unassisted AI content strategies can be effective in SEO
  • We do believe, however, that high-quality AI-assisted content with a human in the loop can perform well in SEO, but we recommend against unassisted low-quality AI content without a human in the loop
  • Full Study: https://lnkd.in/ereSfTzv

Methodology & Data

  • We selected a set of 2.2k keywords across 10 categories with 12k URLs
  • We used originality.ai to measure the prevalence of AI content on each URL and scored URLs as human-written, low AI, high AI, and fully AI
  • Crypto, commerce, finance, and local have the most AI content, which makes sense since these are more descriptive categories that may not benefit as much from personal experience
  • Food, news, travel, and education have the least AI content, which makes sense since they benefit the most from personal experience

Caveats

  • AI detectors are imperfect, and originality.ai is imperfect, so there may be some error in the data, though AI detectors anecdotally tend to overcount AI content vs. undercount
  • We did not control for content quality, and high-quality AI content may perform better than low-quality AI content
  • Human-written content has a cost, whereas AI content costs little to nothing, so the ROI of AI content may still be compelling even if the overall performance is lower

Opinions

  • Google has recently announced guidance around AI content, saying that AI content is OK so long as it’s high-quality
  • I suspect that Google has built an internal AI detection algorithm to detect AI content + low-quality AI content
  • I think that Google is currently penalizing domains and URLs with a high % of AI content
  • Dilemma: If low-quality AI content is effective in SEO, everyone will do this, and Google will become a search engine for LLM responses and thus no longer have a reason to exist - go directly to the LLM
  • Dilemma: If LLMs cannot distinguish AI content from human content, LLMs will use LLM outputs as inputs to themselves, which means there will be an infinite loop of LLM’s training on themselves, which results in an infinite loop of derivatives of derivatives thus converging on a single point of view on everything

Comments

  • If you have data, examples, or learnings about when AI content can be effective in SEO, please feel free to provide these in the comments
  • AI content is a controversial and important subject to evolve how we think about it, please agree or disagree with our opinions in the comments
Ethan Smith™
Add to my network

Location

San Francisco Bay Area