Rethinking SEO in the age of AI | Eli Schwartz (SEO advisor, author)
Eli Schwartz, a growth advisor specializing in SEO and author of Product-Led SEO, discusses the significant shift in SEO due to AI overviews and LLMs. He explains how Google's integration of AI answers is changing search behavior and what businesses must do to succeed in this new paradigm.
Deep Dive Analysis
20 Topic Outline
Initial Thoughts on AI's Impact on SEO
Google's AI Overviews and Search Generative Experience
Impact of AI Overviews on User Search Journeys
Top-of-Funnel vs. Mid-Funnel SEO Strategies
Treating SEO as a Product Problem
Case Studies: Zapier and Tinder's SEO Approaches
Three Steps to Get Started with SEO
When SEO is Not the Right Marketing Channel
Evaluating SEO Investment and Trade-offs
Measuring SEO Success: Conversions vs. Traffic
Understanding the Time Horizon for SEO Results
Leveraging AI in Content Creation for SEO
AI Overviews: Branding Opportunity or Content Theft?
Brand Building as the Foundation for Link Building
Programmatic vs. Editorial SEO Strategies
Key Takeaways from the Google Antitrust Verdict
Debunking Common SEO Myths
Forecasting SEO Success: Top-Down Approach
The Evolving Need for SEO Expertise
TikTok and Instagram as Search Alternatives
7 Key Concepts
SEO as a Product
A mindset where SEO is approached like product development, focusing on understanding user needs from the search channel and creating a valuable experience or asset for them, rather than just a marketing tactic.
Top-of-Funnel SEO
Search queries where users are in the early stages of discovery or curiosity, often using broad keywords. These queries are increasingly being addressed by AI Overviews, reducing the need for users to click through traditional search results.
Mid-Funnel SEO
Search queries where users have a clearer intent and are looking for specific solutions or comparisons, often after initial discovery. This stage is becoming more critical for SEO as AI Overviews guide users to this point.
AI Overviews (formerly SGE)
Google's feature that provides AI-generated summary answers directly at the top of search results, essentially integrating a ChatGPT-like experience into Google Search. This feature aims to provide direct answers, potentially reducing clicks to original sources.
Programmatic SEO
An SEO strategy that involves automatically generating a large number of unique, data-driven pages by combining various data sources and templates. This approach is effective for scaling content that addresses specific user needs at scale, like property listings or product comparisons.
Editorial SEO
The traditional approach to SEO that involves manually writing long-form, unique content, such as blog posts or articles. This method is typically less scalable and may be less effective for queries now handled by AI Overviews or programmatic solutions.
Unstructured Data
Information found in free-form text, like blog posts or articles, which Google's AI Overviews are now capable of summarizing and structuring into direct answers. This contrasts with structured data, which is already organized and easily extractable.
12 Questions Answered
AI Overviews (formerly SGE) are now integrated into Google search, providing AI-generated answers directly. This pushes traditional search results down and means users get answers without clicking links, especially for top-of-funnel queries.
Top-of-funnel searches (broad queries) are increasingly being answered by AI Overviews, reducing clicks to websites. Mid-funnel searches (more specific, intent-driven queries) are where SEO will continue to matter, as users seek deeper information or solutions after initial AI-driven discovery.
Businesses should treat SEO as a product problem, focusing on understanding the user's self-discovery journey and creating product experiences that address mid-funnel needs. This requires collaboration between product and marketing teams.
SEO is often not a good fit for B2B SaaS companies or products with complex sales cycles, high price points, or where the problem isn't actively searched for. If the user journey doesn't lead to an online conversion, other channels like sales or brand marketing may offer better ROI.
First, deeply understand your user and what problems they would search for that your product solves. Second, decide what valuable asset or product experience you need to create for that user. Third, envision and build this product, involving design, engineering, and product management resources.
SEO success should be measured by conversion metrics relevant to the business, such as MQLs (Marketing Qualified Leads), actual dollar conversions, or specific user actions (e.g., phone calls, video views), rather than vanity metrics like top-of-funnel rankings or raw traffic volume.
Yes, AI can be a useful tool for content creation, especially for structured or descriptive content like e-commerce product descriptions, or to assist human editors. However, it should not be used to generate large volumes of low-quality, unhelpful blog posts solely for SEO purposes.
Showing up as a brand mention in an AI Overview indicates strong branding efforts. However, if your content is merely linked, Google might be 'stealing' your content by summarizing it, potentially reducing direct traffic to your site. Companies should focus on building a strong brand that gets mentioned.
Programmatic SEO involves generating many unique pages by combining data sources and templates (e.g., Zillow's property pages, TripAdvisor's hotel pages). Editorial SEO involves manually writing long-form content like blog posts. Programmatic is often more scalable and effective for specific user use cases.
Keyword research tools are often inaccurate, sometimes overestimating or underestimating search volume by many factors, because they rely on proprietary algorithms and estimations rather than direct Google data. They are useful for normalization and understanding user journeys but not for precise forecasting.
Instead of a bottom-up forecast based on unreliable keyword volumes, use a top-down, TAM-based approach. Estimate the total addressable market (e.g., population, demographics), apply relevant filters (e.g., online buyers), and then project market penetration and conversion rates.
While Gen Z uses platforms like TikTok for top-of-funnel discovery and visual content (e.g., travel inspiration), Google Search remains dominant for mid-funnel, intent-driven searches like booking hotels or flights, where users need to take specific actions. Different platforms serve different parts of the user journey.
25 Actionable Insights
1. Treat SEO as a Product
Approach SEO as a product question, with product managers thinking about the user experience and how to position the product for self-discovery journeys, rather than just a marketing challenge.
2. Cultivate User Empathy
Start your SEO process by deeply understanding your target user’s perspective and problems, asking what they would search for to find your product.
3. Validate User Search Intent
Do not invest in SEO if you cannot clearly identify what users would search for to find your product, as SEO is about appealing to that specific user intent.
4. Evaluate Online Conversion Potential
Assess if your product’s buyer journey is primarily online and self-serve; if it requires extensive sales interaction or multiple stakeholders, SEO may not provide a good return on investment for direct conversions.
5. Prioritize Channels by Conversion
If your product’s core conversion motion is sales-led rather than online, other channels like brand ads or influencer campaigns might yield faster and better ROI than SEO.
6. Focus on Mid-Funnel SEO
Shift SEO efforts to target mid-funnel queries, as AI overviews will handle top-of-funnel discovery, leaving opportunities for deeper, more specific searches where users have a clearer sense of what they want.
7. Align SEO with Buyer Journey
Create SEO solutions that directly address mid-funnel problems within the buyer’s journey, ensuring your efforts remain relevant and effective regardless of AI overview changes.
8. Define Clear SEO Conversion Metrics
Clearly define specific business-relevant conversion metrics for your SEO efforts (e.g., MQLs, dollar conversions, phone calls) rather than relying solely on top-of-funnel rankings or traffic, which may not indicate business value.
9. Avoid Traffic for Traffic’s Sake
Do not generate traffic to content that doesn’t convert, as Google doesn’t inherently reward high traffic if it’s not useful or relevant to your business goals.
10. Build SEO as a Product
Envision and build your SEO solution as a product, requiring design, engineering, and product management resources, not just content creation, to serve the SEO user effectively.
11. Design SEO Assets Strategically
After understanding your user, decide what specific SEO asset (e.g., programmatic page, content) you need to create to meet their search intent and solve their problem.
12. Set SEO Milestones
Treat SEO development like a product roadmap, setting clear milestones for ideation, building, and shipping, then track progress against these to assess effectiveness.
13. Favor Programmatic SEO
Generally, programmatic SEO is a more effective and scalable approach than editorial content for many businesses, especially when there’s a clear user use case and data sources.
14. Programmatic SEO for Scale & Use Case
Implement programmatic SEO only when there’s a clear user use case and potential for scale, ensuring the generated pages provide a genuine solution and are not just content for content’s sake.
15. Build Brand for Link Acquisition
Focus on building a strong brand that naturally earns mentions and links, rather than engaging in artificial link-building tactics like buying guest posts on low-authority websites.
16. AI Overviews as Branding
View appearing in AI overviews as a branding exercise; if your brand is mentioned, it signifies successful branding efforts, but merely having your content linked might mean it’s being given away for free.
17. Prioritize Content Usefulness
Focus on creating helpful and useful content, regardless of whether AI is used in its generation, as Google’s primary concern is content quality and relevance, not its origin.
18. Use AI for Product Descriptions
Leverage AI to generate product descriptions for e-commerce sites, as this type of content directly supports product optimization rather than creating generic, less useful blog posts.
19. Follow Google’s SEO Basics
Adhere to Google’s best practices for SEO: build an understandable website, ensure proper internal linking, and create helpful, user-centric content, as these simple steps unlock significant initial success.
20. Assess Technical SEO Needs
Prioritize technical SEO only if you have a massive website with millions of pages; for smaller sites (e.g., a 100-page SaaS tool), it’s less critical and investing heavily in it could be a waste of time.
21. De-emphasize Page Speed Fixes
Avoid overspending on page speed optimizations unless you are in a highly competitive niche where marginal speed differences significantly impact user experience, as it often provides diminishing returns for most sites.
22. SEO for User Action & Business Benefit
Frame your SEO strategy around facilitating user actions that directly benefit your business, recognizing that the aggregate demand for these actions remains constant even if search volumes shift.
23. Use Top-Down SEO Forecasting
For SEO forecasting, especially for new market entries, use a top-down (TAM-based) approach by estimating market size and penetration, as bottom-up keyword-based forecasts are often inaccurate and underestimate potential.
24. Use Keyword Tools for Trends, Not Exacts
Utilize keyword research tools for indicative trends and understanding user search behavior (e.g., spelling variations), but avoid relying on their exact volume numbers for precise forecasting due to their inherent inaccuracies.
25. Develop Consulting Skills (Moonlight)
If aspiring to be a growth advisor, build your consulting skills (sales, closing, retention) by moonlighting while keeping your day job, rather than immediately quitting to become an advisor.
9 Key Quotes
Think of SEO as a product. The product managers are the people that should be thinking about this SEO question because it's a product question.
Eli Schwartz
Transparently, I thought this was going to be an apocalypse. Up until AI overviews, whoever won on that long form piece of content would get that first click. But now that doesn't exist anymore.
Eli Schwartz
If you can't answer the question about what is it that someone's going to do a search on, then don't do SEO.
Eli Schwartz
To a lot of people, SEO is kind of this dark art. It is not a dark art. It is simple.
Eli Schwartz
The biggest myth in SEO is that you could just build links and link building is the secret source and the secret sauce to like growing your SEO footprint. And that's totally wrong because the way most people build links is they buy these guest posts or they pay for links on low authority websites that sort of look like they might have authority.
Eli Schwartz
I don't think there's a chance that chat GBT or perplexity or Claude or any of these other LLM startups... have a chance. They're really unseating Google purely from a quality standpoint.
Eli Schwartz
I think the most important thing to leave people with is that SEO is not dying. There will always be a world where users are requesting their own information.
Eli Schwartz
Google believes that we're in the very early days of LLMs. And even with all the machine learning they do and understanding users, you will always need real user data, which Google has on past searches.
Eli Schwartz
I don't know that any of them can get close to the truth was when I've worked with big companies where there were keywords that I could look at Google search console. Some of them, the tools were overestimating by 10 times. Sometimes they're underestimating by 10 times.
Eli Schwartz
1 Protocols
Getting Started with SEO in the New Paradigm
Eli Schwartz- Be the user: Try to understand who your user is and what problem they would search for that your product solves. If you cannot answer this question, do not do SEO.
- Decide on the asset to create: Based on your understanding of the user, determine what product or experience you need to build (e.g., a programmatic page, a specific tool) that showcases how your product solves their problem.
- Envision and build the product: Treat the SEO solution as a product, requiring design, engineering, and product management resources to oversee the building of the necessary pages or features.