Product-Led SEO
Treat SEO as a product you build for the search user, not content you optimize for a keyword.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 3
- Confidence
- 95%
Eli Schwartz's core reframe: SEO is not a content exercise owned by marketing but a product exercise that needs a PM, design, engineering and user research. The search visitor is a distinct user segment — one who will never hear about you from an ad, a social channel or a trade show — and you build a product surface specifically for their self-discovery journey. The three-step sequence is: be the user, decide the asset, build the product.
Origin
Developed by Eli Schwartz across a decade of SEO work (SurveyMonkey, Zapier, Tinder, Quora, WordPress, Fair) and codified in his book 'Product-Led SEO'. The trigger was a CEO who reduced SEO to 'find keywords, write content, build links — why should I pay you?', which pushed Schwartz to pivot his thinking from tactics to product.
Core principles
- 01The search visitor is a user segment with its own journey, not a traffic number.
- 02If you cannot state what a user would type into a search box to reach your product, you have no SEO product to build.
- 03Inputs to SEO are design, engineering, user research and a PM — not just a writer.
- 04The asset must sit inside the buyer journey, otherwise the traffic goes nowhere.
- 05Customer empathy precedes keyword research.
How to run it
- 1
Be the user
Before touching keywords, work out who the search user actually is and what problem drives them to search. Use the product yourself. Schwartz notes SurveyMonkey employees had accounts but never ran a survey, so they had zero customer empathy for the tool they were marketing.
Pro tip Force yourself and your team to actually use the product as a customer would — onboarding-time usage builds the empathy that keyword tools cannot.
Watch out If you ask 'what would someone search to find this?' and get a blank stare, stop. That is the signal not to do SEO at all.
- 2
Decide the asset
Given that user and that problem, decide what you must create. For Tinder the user was solving loneliness in a new city, so the asset had to be local and global at once — which meant programmatic city pages, not long-form dating content.
Pro tip Let the user's problem dictate programmatic vs editorial rather than picking a format first.
Watch out Do not default to 'write a long blog post'. Nobody converts into Tinder by reading an article on how to fall in love.
- 3
Build it as a product
Resource it like a product: PM oversight, design, engineering, data inputs, user research, a roadmap with milestones. Define what data feeds the page and what the page must look like for that user to act.
Pro tip Run it on a product roadmap with dated milestones (ideate month 1, PRD month 2, ship month 3) so a stall is visible as a missed milestone rather than 'SEO didn't work'.
Watch out If engineering will not be resourced, the plan dies. Schwartz has seen a CTO force a choice between 'the product' and 'this marketing thing' — and the SEO plan simply never shipped.
In the wild
Nobody searched for 'Zapier' because nobody knew it existed — but people did search for whether Gmail connects to Salesforce. Schwartz and Wade Foster built, at scale, a page for every pairing of tools that could be zapped together, showcasing that the connection existed.
→ Became one of the most-copied programmatic SEO successes in tech, creating a flywheel where discovering one integration prompted users to look for the next.
All of Tinder's inbound came from the word 'Tinder'. Working with the head of growth, Schwartz reframed Tinder from a dating app to a loneliness solution — a problem that spikes when you move to a new city — and built programmatic 'online dating in [city]' pages worldwide that surfaced date spots and Tinder as the fix.
→ A mid-funnel SEO surface that survives AI Overviews, because an AI answer can describe the dating scene in Dubai but cannot solve the user's loneliness.
Common mistakes
Building the product first and handing SEO to marketing afterwards
Marketers are then asked to bend an existing product into whatever the search results reward. Nobody ever asked what the searcher wanted, so the fit is fake and the pages do not convert.
Doing SEO to compensate for a bad product
Schwartz describes a health app publishing thousands of AI blog posts while the app itself was awful. Marketing cannot make a product fit a promise the product does not keep.
Is it for you?
Best for
A PM or growth lead at a company with a genuine self-serve search journey who has been handed 'do SEO' and wants to build a durable asset instead of a content mill.
Not ideal for
Sales-led enterprise products with committee buying, where no search result ever ends in a purchase.
From the transcript
“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”
“so going through those steps again it's understand who your user is decide what it is that you need to create for that user and…”
“I think of it as a product which means you need design resources you need engineering resources of course you need a product manager to…”
From the episode
Rethinking SEO in the age of AI
Eli Schwartz (SEO advisor, author)