SEO is not dying but evolving: success now requires treating it as a product-led discipline focused on mid-funnel user journeys, not top-of-funnel keyword rankings.
Eli Schwartz says top-of-funnel discovery searches are being swallowed by AI answers. Until AI overviews, whoever won the long-form piece of content got the first click; now Google or ChatGPT answers in a paragraph and pushes results down, so the user moves on to mid-funnel searches, which is where SEO still matters.
Treat SEO as a product rather than a marketing task. Schwartz argues product managers should own it: understand the user doing a self-discovery journey who won't come from social, ads or trade shows, then build the experience and positioning that meets them in the mid-funnel, with product and marketing collaborating closely.
Step one, which almost everyone misses, is to be the user and understand what they would actually search for. Step two is deciding what asset to create, such as programmatic pages like Tinder's city pages. Step three is building it as a product, with design, engineering, a product manager and user research.
Mostly not, according to Schwartz. Tools like Mixpanel are expensive, integrated company-wide and not bought from a search click, and Google Cloud decisions go through committees against Amazon and Microsoft. People rarely search for a problem they don't know they have, so awareness is usually a brand challenge and buying is a sales motion.
SEO is not free. A typical agency costs upwards of $10,000 a month, or $120,000 a year, before adding a CMS, engineers, design and content. Schwartz suggests comparing whether a million dollars in SEO would pay back sooner than the same money spent on brand ads, influencers or paid marketing on Meta and Google.
Track a conversion metric that matters to the business: MQLs, dollar conversions, page views for media companies, or links for a startup. Schwartz calls reporting SEO traffic alone in investor decks a total waste of time, and says rankings matter only if someone like investors cares about them. Traffic that doesn't convert is worthless.
It depends on what you build. For big brands like Tinder, results come quickly once pages exist, while smaller companies may wait many months for Google to notice; Zapier took a couple of years. Schwartz recommends a product roadmap with milestones, such as pages indexed after the first month, to judge progress.
It depends on the user. Programmatic SEO combines data sources into pages at scale, like TripAdvisor's property pages or Zillow's home values, and is generally the better path when there is scale and a real use case. Editorial long-form content for 500 million properties would be cost prohibitive and not what users want.
AI is a tool, not a solution. Using it to write whole blog posts just produces the same worthless content people once bought on Upwork for $50, now for free. It is useful for adding product descriptions, titles and features to pages, as e-commerce sites like Macy's rely on product pages rather than fluff content.
Schwartz read the 286-page verdict. It states Google has 98% of mobile searches and shows how default distribution deals, like revenue share with Apple, drove its success; Neva failed despite a better product. Bing offered itself to Apple for free and was refused, so he doubts ChatGPT or Perplexity will unseat Google.