Lenny's PodcastEthan Smith: How Webflow saw 6x conversion from LLM traffic
In LLM answers the breadth of citations beats classic page rank; Reddit, YouTube, and Dotdash listicles feed the model, and Webflow saw 6x LLM conversion.
CHAPTERS
- 0:00 – 4:41
Why AEO matters now: getting recommended inside ChatGPT
Lenny and Ethan frame the episode around the rise of Answer Engine Optimization (AEO): how to get products and content recommended by LLMs like ChatGPT. They preview why citations and clickability in AI answers are changing acquisition, and tease that AEO can be a meaningful new growth lever.
- •What AEO is and why people suddenly care
- •LLMs summarize multiple sources, changing how “winning” works
- •ChatGPT already driving meaningful referral traffic for publishers
- •Early claim: AEO can be influenced/optimized (but not via spam)
- 4:41 – 6:18
The biggest SEO shifts: from anti-spam (Panda) to AI-overview/search summarization
Ethan places AEO in historical context: the largest SEO shift was Google’s move from spam-friendly to anti-spam algorithms; AEO is the next major wave. He argues this is still “search,” but with summarization and new inputs, not a total reset of everything that worked before.
- •Ethan’s SEO background since 2007 (programmatic/commerce SEO)
- •Panda-era change: mass autogenerated pages stopped working
- •AEO is the second-biggest shift: search + summarization + new inputs
- •Core takeaway: not everything is different; fundamentals still apply
- 6:18 – 9:48
AEO vs GEO: definitions and what you’re actually optimizing
They define AEO (Answer Engine Optimization) and GEO (Generative Engine Optimization), with Ethan arguing they largely refer to the same practice today. He prefers “answer” as a clearer, narrower framing focused on surfacing in responses, not all generative media.
- •AEO and GEO used interchangeably in practice
- •Why “answer engine” is more precise than “generative engine”
- •Goal: show up as an answer inside LLM interfaces
- •Terminology will likely settle based on industry adoption
- 9:48 – 12:01
Can you really influence AI answers? Impact examples and what changes in the “head”
Ethan explains how AEO differs from classic blue-link SEO at the top of the funnel: LLMs use multiple citations, so being mentioned more often across sources can matter more than being the #1 cited URL. Using Webflow as an example, he shares what’s working: traditional SEO plus YouTube and Reddit citation wins.
- •LLMs summarize multiple citations; “#1 link” isn’t the whole game
- •Winning often correlates with most-cited brand across sources
- •Webflow example: traditional SEO still helps; plus YouTube + Reddit
- •AEO opens faster paths for visibility than domain-authority-driven SEO
- 12:01 – 14:35
Early-stage startups can win fast: citations tomorrow + the bigger conversational long tail
Ethan contrasts early-stage SEO (hard without authority) with early-stage AEO (possible quickly through fresh citations). He also introduces the “tail” difference: prompts are longer and more conversational, creating many new, ultra-specific questions that never existed in keyword search volume tools.
- •SEO early-stage often not worth it due to domain authority constraints
- •AEO can move quickly via new citations (Reddit, blogs, videos)
- •Chat prompts are longer (bigger long tail than search)
- •Opportunity: answer never-before-asked questions in niche categories
- 14:35 – 15:50
Lead quality from LLMs: why conversions can be dramatically higher
They dig into whether AEO traffic is valuable, and Ethan claims it can be significantly more qualified than search traffic. He shares a striking Webflow data point showing much higher conversion rates from LLM-referred visitors, likely due to intent built through conversational narrowing.
- •LLM traffic can convert far better than Google search traffic
- •Webflow example: reported 6x conversion rate vs search
- •Conversations build intent through multiple follow-ups
- •Being “the answer” compounds trust and readiness to act
- 15:50 – 20:13
On-site vs off-site AEO: citations, content depth, and why Reddit is heavily weighted
Ethan breaks AEO work into on-site (traditional SEO + feature/use-case depth) and off-site (earning mentions in common citation sources). They discuss why Reddit is disproportionately cited and how to approach it without spam—using transparent, helpful participation instead of fake accounts.
- •On-site: pages must answer many follow-up questions (features, integrations, use cases)
- •Off-site: win citations across video, UGC, affiliates, blogs, major publishers
- •Reddit is highly cited; spam attempts get banned/deleted
- •Effective Reddit play: disclose identity/employer, provide genuinely useful answers
- 20:13 – 21:54
How LLMs use citations: RAG vs the core model (and what’s actually controllable)
They clarify that most actionable AEO affects retrieval (RAG) rather than the base model’s training. Ethan explains the two-layer system—core next-word prediction vs retrieval-augmented generation—and argues RAG is where marketers can realistically influence visibility and timing.
- •Two systems: core model training vs RAG (search + summarization)
- •AEO tactics mostly influence RAG citations, not base training
- •Core-model influence is slow and impractical for most companies
- •Platforms tune which sources to trust (similar to Google ranking choices)
- 21:54 – 24:59
Core principles to win at AEO: topics, question research, and citation categories
Ethan shares key mental models: optimize around topics (each page covers many questions), answer follow-ups comprehensively, and do dedicated question research without a clean ‘truth set’ like Google keyword volume. He recommends mining sales/support questions and mapping citation types to tailored strategies.
- •AEO is still search-like: LLM + RAG summarizing results
- •Topic-based optimization: pages target hundreds/thousands of questions
- •Question research is harder without platform volume data
- •Mine sales calls, support tickets, Reddit/UGC for real question demand
- •Segment citations by type (site, video, UGC, affiliates, blogs, major media)
- 24:59 – 28:53
Avoiding hyper-SEO’d sameness: originality, information gain, and “typicality” detection
They discuss why modern SEO content often feels derivative and how AI/search engines may evolve to reward originality. Ethan outlines signals like information gain, atypicality (not just rewriting SERP averages), and domain expertise/original research as long-term defensible advantages.
- •Problem: content scoring tools incentivize everyone to rewrite the same SERP patterns
- •Most pages drive little impact; pushes low-cost, low-expertise content production
- •Potential ranking signals: information gain, atypicality, expertise indicators
- •Best content: original research + credible domain expertise, explicitly stated
- 28:53 – 34:03
Actionable AEO playbook: pick questions, track share of voice, create pages, earn citations
Ethan lays out an execution plan: derive target questions from paid/search data, track them with answer trackers, analyze current citations, and build the page types that win. Off-site, he suggests options ranging from paid affiliates to YouTube/Vimeo and carefully executed Reddit participation—then validate through experiments.
- •Turn money keywords (yours + competitors’) into questions using LLMs
- •Use answer tracking to monitor frequency and average rank across runs
- •Study citation sources and mimic winning page formats (listicles, category pages, etc.)
- •Off-site levers: affiliates (easy but costly), video (underused in B2B), Reddit (high leverage but community-governed)
- •Build a team: SEO + someone capable of off-site/community/video execution
- 34:03 – 38:25
Measurement and experimentation: control groups, variance, and reproducibility
They go deeper on how to measure AEO where answers vary across runs, surfaces, and prompt variants. Ethan recommends true experimentation with control groups across question sets, repeated trials for reproducibility, and caution against ‘best practices’ that were never tested.
- •Answers vary by run, prompt variant, and platform (ChatGPT vs Perplexity vs Gemini, etc.)
- •Track “share of voice” across surfaces rather than single rankings
- •Experiment design: split questions into test/control; intervene only on test group
- •Need control due to baseline volatility and overall adoption growth
- •Reproducibility matters—avoid attributing wins to the wrong change
- 38:25 – 41:46
Segment-specific AEO: B2B vs commerce vs early-stage + attribution pitfalls
Ethan explains how AEO tactics and measurement differ by business type. B2B often lacks clickable modules, making last-touch attribution unreliable and requiring post-conversion self-reporting; commerce/local have richer clickable cards where schema/reviews matter. For early-stage, he recommends focusing on citations and ultra-long-tail questions rather than mid-funnel SEO plays.
- •Citations differ by category: B2B sources vs commerce magazines vs local/reviews sites
- •B2B tracking is harder: fewer clickable answers; multi-touch journeys; need surveys/assist metrics
- •Commerce/local: clickable shopping cards; schema and reviews become key inputs
- •Attribution confusion: users may later search brand/direct, masking LLM influence
- •Early-stage strategy: prioritize citations + long tail, skip heavy traditional SEO builds
- 41:46 – 51:49
Indexing vs training: should you let AI crawl your content?
Lenny asks whether it’s ‘good’ for LLMs to ingest publisher content. Ethan argues you’re in the game regardless; blocking indexing means losing visibility to competitors. He suggests selectively blocking training while allowing indexing via robots.txt/user-agent controls.
- •You can’t opt out of the market dynamics—competitors will show up instead
- •Allow indexing for discoverability; consider blocking training if desired
- •Use separate bots/user agents to distinguish training vs indexing permissions
- •Practical approach: configure robots.txt (and tooling) rather than blanket bans
- 51:49 – 58:55
AI-generated content: what works, what doesn’t, and the risk of infinite derivatives
Ethan shares research suggesting fully automated AI content (no human-in-the-loop) underperforms in both Google and ChatGPT citations, while AI-assisted editing is the real future. He warns that if AI derivatives dominated citations, systems risk ‘model collapse’ and convergence on single bland opinions—breaking the wisdom-of-crowds value of retrieval.
- •Distinction: AI-assisted content (good) vs 100% AI-generated autopublish (bad)
- •Study claim: only ~10–12% of top results/citations are AI-generated; most are human
- •Broader web trend: AI-generated pages may now outnumber human-created
- •If derivatives ranked, search becomes “AI summarizing AI,” incentivizing infinite loops
- •Risk: collapse of diversity—converging to one opinion; degraded information ecosystem
- 58:55 – 1:03:19
The future: convergence of LLMs and search + help-center optimization for the long tail
Ethan predicts LLM and search experiences will converge (AI overviews, maps, shopping modules). He ends with an underused AEO lever: help centers—where detailed feature/integration questions live—and offers concrete optimization steps like moving from subdomain to subdirectory, improving internal linking, and writing articles for obscure-but-real tail use cases.
- •Search + LLM UX converging (AI overviews, shopping, maps)
- •Anti-spam priorities: prevent scraped/AI-derivative loops
- •Help center is a powerful long-tail asset for ‘does it do X?’ questions
- •Tactics: move help center to subdirectory; strengthen cross-linking
- •Fill the tail using sales/support questions; consider community Q&A to scale coverage
- 1:03:19 – 1:11:55
Lightning round: books, media tastes, gear, and Ethan’s proudest SEO win
In the closing segment, Ethan shares recommended books centered on psychology and measurement, plus personal media preferences and favorite recent purchases. He also highlights a standout SEO win (MasterClass ‘butter lettuce’) and closes with where to find his work and how listeners can help.
- •Book recs: Emotional Intelligence; Cialdini on persuasion; How to Measure Anything
- •Media: intense sports/UFC and climbing documentaries (flow + craft)
- •Favorite products: mirrorless camera + Shure mic for high-quality calls
- •Proudest SEO story: MasterClass ranking success (butter lettuce)
- •Where to follow: LinkedIn + Graphite research; ask: share AEO experiment results + comment thoughtfully
