Search Everywhere Optimization

Why the framework exists

The web did not get replaced. It multiplied. In 2026 a single query hits Google, but it also hits ChatGPT, Claude, Gemini, Perplexity, Google Lens, Siri, Alexa, the App Store, the Google Business panel, an AI agent booking a flight, and a knowledge graph feeding half of them. Optimizing a page for one of those channels is a partial answer. Search Everywhere Optimization is the framework I use to cover all of them without doing the work thirteen times.

The problem the framework solves is straightforward. Most SEO agencies still ship a keyword list, a title tag audit, and a link-building plan. That work is real, but it maps to one of the thirteen surfaces I track. The other twelve stay unclaimed. When AI Overviews started answering questions above the classic ten blue links, sites that had optimized only for Google lost the click even when they still held the rank. The framework assumes that gap keeps widening.

Search Everywhere Optimization is a working document, not a theory. It sits behind the case studies on this site and behind the article system on fredericksona.com itself.

The thirteen surfaces

The framework tracks thirteen distinct discovery surfaces. Each has its own signals, its own feedback loop, and its own rate of change. The point of naming them is to stop conflating them.

• SEO. Classic Google and Bing results.
• AEO. Answer Engine Optimization. AI Overviews, Perplexity, Bing Chat.
• GEO. Generative Engine Optimization. Citations inside ChatGPT, Claude, Gemini.
• AAO. Agent Access Optimization. What autonomous agents can do on your site, at what latency, with what schemas.
• VSO. Voice search. Siri, Alexa, Google Assistant.
• VxSO. Visual search. Google Lens, Pinterest Lens, image reverse-search.
• ASO. App Store Optimization. Apple App Store and Google Play.
• KGO. Knowledge Graph Optimization. Wikidata, Google Knowledge Panels, entity records.
• LSO. Local search. Maps, "near me" queries, Google Business Profile.
• CWV. Core Web Vitals. Speed as a ranking signal, not just a UX signal.
• E-E-A-T. Experience, expertise, authoritativeness, trust. Signals for both classic and AI surfaces.
• GLOBO. International and locale ranking. hreflang, currency-aware pipelines, market-specific entity records.
• Web3. ENS, Farcaster, decentralized identity. The smallest surface today and the least stable, kept in the framework as a first-mover position.

The map

The surfaces cluster into four priority tiers by return per hour of work, not by importance.

Priority tiers

The tiers order surfaces by return per hour of work, not by importance. Every surface matters. What changes between tiers is the cost of shipping and how quickly the improvement compounds.

Tier I is the table stakes. SEO, AEO, GEO, and Core Web Vitals share most of the underlying assets. A clean sitemap, canonical tags, JSON-LD, an OpenGraph pass, an FAQ schema, and a performance budget under 2.5 second LCP cover the classic search stack and most of what AI Overviews and LLM citations look for. This is where I start every engagement.

Tier II compounds. E-E-A-T markup (Organization schema, author schema, freshness stamps, first-party evidence) is cheap to ship once and pays back on every future page. Visual search (alt text plus ImageObject schema on every meaningful image) is often the largest source of Gen Z discovery for consumer brands. Voice search sits next to AEO: the same speakable FAQ block usually covers both.

Tier III is asymmetric. Agent Access Optimization is my favorite surface right now. llms.txt is roughly where robots.txt was in 1998, and almost no sites ship a proper MCP server yet. Knowledge Graph Optimization (a filed Wikidata entity, the sameAs graph, entity disambiguation) is slower to land but powers a growing share of both classic and AI answers.

Tier IV is situational. App Store Optimization only applies if there is an app. Local search only applies if there is a physical presence. International only applies at the second market. Web3 is the smallest and least stable surface, worth the twenty minutes it takes to register an ENS record and link a Farcaster handle, not much more.

The tiering is not a rank. It is a sequencing tool that keeps a project from spending week three on decentralized identity while the sitemap is still broken.

How the framework applies across industries

The framework is industry-agnostic because the surfaces are. What changes is which tiers dominate the mix.

Print and brand shop. Tier I dominates. Product SEO, Google Business Profile, hreflang for a second-market store. VxSO matters because print buyers search image-first. AAO matters for the price calculator, which an agent should be able to hit directly through a documented endpoint.

Career platform. E-E-A-T is disproportionately important because trust decides whether someone hands over their resume. GEO matters because LLMs are already the second place people look for job advice. AAO is the moat. Agents comparing job platforms is a plausible near-term behavior.

Personal site. SEO plus GEO plus KGO. The job of the site is to get the author cited by name in LLM answers about the domains the author writes about. Every article ships with an author schema, an entity link, and a TL;DR block that AI Overviews can lift verbatim.

Physical business. LSO carries the weight. Tier I still ships, but Google Business Profile plus review velocity plus local schema is where the traffic actually comes from.

The framework does not tell you which industry you are in. It tells you, once you know, which surfaces to weigh most and which to skip until the business hits a specific gate.

The unifying principle

Every surface rewards roughly the same underlying behavior. Ship these five once and the framework mostly runs on its own.

• Structured truth about what the page is. Schemas everywhere the surface can read.
• Direct answers to actual questions. TL;DR blocks, FAQs, spec tables, price tables.
• Content that does not exist elsewhere. Original evidence, first-party numbers, an unusual point of view.
• Speed. CWV became a ranking signal, and now agents time out on slow sites.
• Linked identity. sameAs across the web, social, Wikidata, ENS, Farcaster.

That is the actual point of the framework. Thirteen surfaces stop looking like thirteen different projects and start looking like one project with thirteen dashboards.

Scoring a site against the framework

Before I touch a project, I score the site against the thirteen surfaces on a five-point axis.

• 0. Not shipped. The signal does not exist on the site.
• 1. Partial. Some pages carry it, most do not.
• 2. Consistent. Every page ships it, but the implementation is minimum-viable.
• 3. Considered. The implementation is thought through and the site owns a specific answer for each variant of the surface.
• 4. Compounding. The signal is producing measurable growth in referrals from that surface.

The score is a working document, not a report card. Its job is to tell me which surfaces are cheapest to move next quarter. A surface at 0 or 1 with a low cost-to-ship goes to the top of the sprint. A surface at 3 with a slow feedback loop stays where it is until a compounding event unlocks it.

Most projects I audit sit at 2 for SEO, 1 for AEO, 0 or 1 for GEO, 1 for CWV, and 0 for everything else. The delta between what most teams ship and what the framework wants is not a matter of talent. It is a matter of naming the surfaces so a team can hold each one in mind at once.

What the framework does not do

A framework this broad has real limitations, and the document says so at the front.

It does not tell you what to write about. Topic strategy sits in the content strategy document. Search Everywhere Optimization tells you, given a topic, what shape to publish it in so the piece is legible to thirteen surfaces at once.

It does not give you an off-site link-building strategy. Off-site signals still matter. The framework covers the earned-discovery side of the equation and cross-references a separate playbook for the outreach side.

It does not replace judgment. When two surfaces disagree on how a page should be structured, the framework does not tell you which one wins. That is a call the strategist makes based on what the business needs from that page.