Product · Engineering · Information architecture
Search Everywhere Optimization
A practical framework for making knowledge understandable, reliable, and discoverable across search engines, AI systems, applications, and emerging agent interfaces.
Core premisePages are presentations. Knowledge Objects are the unit of optimization.
- Old model
Search once ended at the page
Search used to follow a familiar path. A person entered a query, a search engine ranked pages, and the person selected a result. That path still matters, but it is no longer the only way people find information.
- New reality
Answers now arrive before websites
People now ask questions through search engines, AI assistants, answer engines, social platforms, internal knowledge tools, and products that combine several sources into one response. In many of these experiences, the user sees an answer before seeing a webpage. The webpage becomes one possible source, not always the final destination.
- Core problem
Readable does not always mean interpretable
This changes the work. A page can be accurate and useful to a person, yet remain difficult for a machine to interpret. A current value can appear beside an old value with no clear timestamp. A calculated metric can look like a reported metric. A source can be visible to a reader but unclear to a retrieval system. These failures cannot be solved by copywriting alone.
- Framework response
Treat discoverability as a system
Search Everywhere Optimization treats discoverability as a product and information system problem. It asks teams to define the knowledge they publish, make its context explicit, expose it through useful presentations, observe how external systems use it, and improve the source system when interpretation fails.
On this pageSystem⌄
The framework in one view.
The framework begins with source knowledge, not with a page or a channel. Each component has a distinct role, but they operate as one system.
Five parts, one system.
Search Everywhere Optimization remains the primary framework. The five parts below explain how the framework works and how teams can apply it.
Knowledge Objects
The stable unit beneath a page, API response, table, product card, or generated answer. A Knowledge Object connects an entity with its attributes, relationships, source, time context, and rules.
Core question: What knowledge should remain consistent across every presentation?
Knowledge Integrity
The conditions that make a Knowledge Object dependable. Identity, freshness, provenance, calculation policy, ownership, and failure handling require explicit treatment.
Core question: What must be true before this knowledge can be trusted?
Retrieval Pipeline
An analytical model for understanding where discoverability succeeds or fails, from candidate discovery through source selection, interpretation, answer use, and attribution.
Core question: At which stage does useful knowledge disappear or lose meaning?
Feedback Loops
A method for turning observed search and AI behavior into product improvements. The goal is not to chase one interface. The goal is to locate structural gaps and improve the source system.
Core question: What should change after we observe an inaccurate, stale, or missing answer?
Search Everywhere Canvas
A one page planning tool that connects user questions, priority entities, Knowledge Objects, integrity requirements, presentation surfaces, evaluation queries, owners, and success measures.
Core question: How will a team turn the framework into an operating plan?
What changes in practice.
| Page first approach | Knowledge first approach |
|---|---|
| Start with a keyword or page idea | Start with a user question and the entity involved |
| Write each page as a separate asset | Define reusable knowledge before selecting presentations |
| Treat structured data as an add-on | Generate structured representations from the same source model |
| Measure rankings and clicks | Also measure discovery, interpretation, freshness, attribution, and business value |
| Fix visibility by changing copy | Identify whether the failure is in data, identity, context, presentation, access, or evaluation |
| Optimize for one search surface | Keep knowledge consistent while testing several discovery surfaces |
Principles.
Knowledge before pages
A page is one interface. The underlying knowledge should be defined before the interface is designed.
Entities before keywords
Keywords show how people ask. Entities establish what the answer is about.
Context travels with the fact
A value without its unit, date, source, or definition can become misleading when extracted.
Source systems matter
Better markup cannot rescue contradictory, stale, or weak source data.
Observe without guessing
External systems are not fully visible. Use controlled evaluation, document observations, and avoid claims about proprietary internals.
Improve the system
Repeated manual edits are a warning. Durable gains usually require better templates, rules, pipelines, or ownership.
Compact vocabulary.
- Search Everywhere Optimization
- The overall framework for designing and improving knowledge across modern discovery surfaces.
- Knowledge Object
- A reusable representation of an entity, attribute, event, metric, relationship, or concept with enough context to remain meaningful across presentations.
- Knowledge Integrity
- The degree to which a Knowledge Object is accurate, current, attributable, consistent, owned, and safe to use.
- Retrieval Pipeline
- An analytical model for examining how information moves from candidate discovery to answer use and attribution.
- Feedback Loop
- A repeatable process for observing external behavior, diagnosing failures, improving source knowledge, and retesting.
- Search Everywhere Canvas
- A planning tool that converts the framework into decisions, owners, measures, and an improvement cadence.
Framework Chapters.
Each chapter focuses on one part of the framework and introduces one practical model.
Knowledge Objects
The foundational unit beneath pages and interfaces
↗︎02Knowledge Integrity
The rules that make knowledge dependable
↗︎03Retrieval Pipeline
Where discoverability succeeds or fails
↗︎04Feedback Loops
How teams observe, diagnose, and improve
↗︎05Search Everywhere Canvas
A one page implementation tool
↗︎