A website strategy does not start with a page, a post, or a keyword list. It starts with the business: what it sells, how it delivers, who buys, what the existing domain already holds, and how every page should relate to the rest of the site.
That sounds obvious until you watch a website plan arrive as a list of keywords. A keyword list knows what people typed. It does not know which service the business delivers, which product category has enough inventory, which old page already owns the intent, or which claim needs a source before it reaches a public page.

Domain-Wide Analysis solves that problem for the whole domain, not for one page. It turns a business and its existing site into one set of connected decisions: what the site represents, which sections and pages need to exist, what each page must do, how the pages connect, and what still needs evidence. Content is one output of that strategy. The navigation, URL plan, redirects, internal links, and work list for every team that touches the site are outputs too.
What does “done” mean for a Domain-Wide Analysis?
Done means the business has one map of its whole website that every team can work from. That map answers six questions at domain level. What does the site represent? Which existing URLs stay, change, merge, or go? Which sections and pages need to exist, and under which parent? What job does each page own? Which locations, categories, and filters earn their own URL? Which facts still need approval?
If those questions have no answer, the uncertainty does not disappear. It moves into the live website. Two pages compete for the same service, a category description repeats across product pages, a location page goes live because a city exists, and a trust claim gets published because nobody recorded the proof.
I use a simple test. Give the analysis to the business owner, a writer, a designer, a developer, and a reviewer. If each person can make compatible decisions from the same source, the analysis is ready. If each person needs to ask what the business “really means,” the analysis is still a pile of notes.
Why does the business definition come before keyword research?
A business model controls page architecture. A direct-service business, a marketplace, a catalogue retailer, and a business with recurring contracts may all target similar words, yet each needs a different site structure, conversion path, proof set, and internal-link model.
Start with the business facts that change a page decision: the core offer, supporting offers, delivery scope, audiences, commercial goal, operational limits, proof assets, and the person who can confirm a claim. The facts are not a formality. They decide whether an idea becomes a service page, collection page, product page, guide, policy page, section, or no page at all.
Consider two businesses. One offers a main service and a related service. The related service needs its own page only when it has a distinct buyer, delivery process, proof requirement, or commercial action. If it shares all four with the main service, a section or a supporting resource may serve the reader better than another money page.
An ecommerce site has a similar decision with categories and products. A product family needs its own collection page when shoppers compare items within that family and the store has enough useful inventory to support the choice. A product attribute, such as material, compatibility, size, or condition, may help a shopper filter or compare, while still lacking a reason to become an indexable landing page.
The business definition prevents the page map from becoming an organised list of guesses. It makes the analyst ask a better question: what does this business need the site to explain before a buyer can make a decision?
Why must the existing domain be inventoried before new pages are planned?
Every existing URL already carries a history. It may own an intent, hold internal links, receive impressions, contain useful content, confuse the site structure, or duplicate another page. A new plan that ignores that inventory usually creates more work than it removes.
The inventory should cover commercial pages, information pages, categories, products, filters, brand routes, policy pages, and any other URLs the platform produces. Each item needs a deliberate action: keep, update, merge, redirect, retire, or review. “Leave it for now” is sometimes a valid temporary decision, but it is not an analysis result.

The difference between update and merge matters. Update means the page owns a useful job and needs stronger coverage, proof, links, or structure. Merge means two URLs are competing for the same job, so the team chooses one home for the entity and moves useful material into it.
This is where teams often create duplicate content without intending to. A page called “Service A” and another called “Service A Solutions” may look different in a spreadsheet. If they answer the same commercial question for the same audience, the search engine and the buyer still meet two versions of the same page.
For ecommerce, the same failure appears in a different costume. A broad collection, a brand collection, a parameterised filter, and a buyer guide can all approach the same product family. The answer is not to paste more words onto all four. The answer is to state which one owns comparison, which one owns a specific product, which one helps users filter, and which one serves an earlier information need.
What does competitor analysis need to produce?
Competitor analysis should produce questions, not a copying plan. I study how comparable sites separate services, collections, product types, trust material, locations, proof, process, filters, and support content. Then I ask which of those decisions match the business in front of me.
If several competitors give a related service its own page, I inspect the condition behind the pattern. Does the service have a different transaction? Does it require different proof? Does it answer a different buyer question? Does the business deliver it as a separate offer? The page becomes a candidate only when the business can support the same logic.
The same discipline applies to catalogue structure. A competitor may expose many filters, brand routes, and subcollections. That may reveal how shoppers compare the products. It does not prove every filter deserves a page that search engines should crawl and index.
This is where market sheep parade thinking produces pages that nobody can defend later. The pattern becomes “other sites have it,” then a copywriter receives the task, then the team spends six months asking why the page has no distinct query, conversion, or internal-link role.
Competitor work is useful when it creates a verified decision. It is useless when it converts somebody else’s navigation into your own without checking the business model beneath it.
How does Framenet stop a website from becoming a topic pile?
Framenet is the point where the analysis identifies the business’s central entity and the entities that surround it. It records the main subject, the context that creates demand, the search intent, the transactional intent, related entities, information nodes, trust factors, and audiences.
The central entity is the main thing the page represents. A supporting entity explains a related service, category, product type, process, audience, place, attribute, proof item, or next step. The distinction gives the site a centre instead of allowing every keyword to behave like a new topic with a new URL.
Think about a main service page. The service is the central entity. Its process, qualification criteria, common use cases, proof, price factors, and related resources are supporting entities or supporting information. They deepen the service page because they help the reader evaluate the service. A second commercial service is different. It may need its own page because it carries a separate transaction and buyer journey.
The same model applies to ecommerce. A collection page can carry a product category as its central entity. Products, product families, use cases, attributes, accessories, delivery, returns, warranty, and buying guidance become supporting entities. The collection page helps a shopper compare the set. The product page helps the shopper evaluate one item. The returns page explains a policy. They share a commercial environment, but they do not perform the same job.
Framenet gives me four rules to apply before any page is built or changed:
- A money page carries one central entity.
- Every supporting entity has a defined home, whether that home is a page, a section, a filter, a product record, a policy, or an internal link.
- Related entities receive links that explain their relationship to the reader.
- A new service, category, market, or product family updates the map before it generates new pages.
The fourth rule saves a lot of cleanup. Adding pages before updating the map is how a site develops rooms that do not connect to the hallway. The copy may read well in isolation. The domain still has no idea where the page belongs.
How does structure turn the entity map into a usable website?
The site structure decides parent and child relationships. It separates main navigation from supporting navigation, commercial paths from resources, proof from policy, and page depth from mere page existence. Without that structure, a Page Map becomes a list of URLs that happen to live on the same domain.
For a service business, the structure may place core commercial services beneath a Services parent. Process, proof, locations, and resources then connect to the right service relationship. A location route belongs under a service only when the service-location combination has a reason to exist beyond swapping a place name into the same paragraph.
For ecommerce, the hierarchy may move from a broad category to a product family, then to a product or configuration. A buyer guide can sit beside that path and link into it. Delivery, returns, warranty, help, and contact pages support the transaction without pretending to be category pages.
Navigation and internal links need different jobs. Navigation shows the routes buyers need frequently. Internal links explain context. A product page links to a compatible accessory because the relationship helps a buyer. A service page links to a process page because the reader may need to understand how delivery works before requesting a consultation.

The footer also has a job. It gives repeated access to trust, policy, support, and business-identity pages. It should not become an attic where unrelated pages go to avoid being deleted.
How does the Page Map decide what each page owns?
The Page Map is where the earlier decisions become production objects. Each row should state a page name, page type, current URL or planned URL, primary query role, parent relationship, verification state, and any important conditions that affect publication.
A row is not an instruction to publish. It is a candidate that must survive the logic around it. A service-location candidate still needs local evidence. A collection page still needs enough product coverage. A filter route still needs a decision about its search role. A guide still needs a distinct information need rather than a title copied from a search export.
The Page Map becomes useful when each row describes one job. A commercial service page explains the offer and helps a buyer take action. A collection page helps shoppers compare a category. A product page explains one item. A guide answers a pre-purchase or ownership question. A policy page states an approved rule. A filter helps people narrow a category. One URL should not try to become all six at once.
| Page role | What it owns | What it should not own |
|---|---|---|
| Commercial service page | Service scope, buyer qualification, process, proof, and conversion path | A separate commercial service with a different transaction |
| Collection page | Comparison across a relevant product set | Product-level claims for every item in the catalogue |
| Product page | One item’s specifications, availability, condition, compatibility, and purchase path | Broad category comparison that belongs higher in the hierarchy |
| Information guide | A distinct question before or after a purchase | A generic sales page with headings added around it |
| Location page | A supportable service-location relationship | A duplicate of the parent service page with a new city name |
| Filter route | On-site narrowing by a meaningful attribute | An automatic indexable page for every parameter combination |
This is where “does this deserve a page?” becomes a real operating question. It forces the team to compare the page’s entity, intent, audience, evidence, place in the structure, and next action against the pages that already exist.
The next post will focus on that Page Map decision in detail. This post establishes the conditions that make the map trustworthy enough to inspect.
Why does keyword research follow page purpose instead of creating it?
Keyword research tells me how people phrase an information or buying need. It does not understand the business model, product availability, delivery limits, existing URL history, or internal-link structure. That is why it supports a page decision rather than creating the decision by itself.
Once a page has a defined job, research assigns the query set that belongs there. A primary query gives the page its main retrieval context. Secondary queries identify related questions, attributes, qualifiers, and language that the page must address where they fit.
The page role controls how the research is used. A commercial page needs terms that help a buyer assess the offer. A collection page needs category language and selection criteria. A product page needs item-specific language and verified product facts. A guide needs informational questions that lead naturally toward the relevant commercial or category page.
The mistake is creating a separate URL for each phrase variation. A collection of related phrases may belong on one strong page because the user still wants the same outcome. A phrase can also belong in a section, a filter label, a comparison row, an internal link, or no page at all. The search export will not make that distinction for you, which is a little inconvenient but better than letting a CSV run the architecture.
Why do service geography and ecommerce filters need their own decisions?
Service sites and ecommerce sites share the same domain-analysis logic, but their structural risks differ. A service site can overproduce location pages. An ecommerce site can overproduce filter, category, and product routes. Both problems begin when a technical possibility gets mistaken for a page decision.
For a service business, a nearby city is only a geographic fact. A location page needs more: real coverage, a useful relationship to the service, local proof where appropriate, and a distinct page purpose. Distance alone is not proof. Search volume alone is not proof either.
For ecommerce, an attribute is only product data until the site decides what job it serves. It may support a filter, an item specification, a comparison table, a buyer guide, or a collection section. A separate landing page needs its own buyer task, product set, and evidence that the page will help rather than repeat the category above it.
The relevant question is consistent across both models. Does this variation change the reader’s decision enough to earn a separate page? If the answer is no, the variation may still matter. It just belongs in a smaller structural unit than a new URL.

How do unresolved questions protect the strategy?
Every serious analysis finds unknowns. A service business may need to confirm scope, delivery boundaries, proof ownership, policy language, or the accuracy of an old claim. An ecommerce business may need to confirm product data, availability rules, compatibility, condition standards, delivery terms, warranty language, return conditions, or category membership.
Those questions need a visible record. Each one should show the question, the reference that raised it, the person or team who owns the answer, and the action needed to close it. A website strategy cannot rest on an unknown, and nobody downstream should turn one into a fact because a page template has an empty slot.
Action items turn analysis into work. They assign an owner, priority, description, and evidence reference. “Collect proof for the main service” is an action. “Confirm the category rule for this product family” is an action. “Make it sound more trustworthy” is a sentence that should be removed from the project board immediately.
Issue tracking adds the reason. A duplicate service URL creates overlap. Missing product data blocks a specification claim. A location route with no proof risks thin copy. A filter with no unique buyer task creates an unnecessary crawl path. Once the issue has a clear explanation, the team can fix the cause instead of building around it.
What does the business receive after Domain-Wide Analysis?
The business receives one website strategy instead of a keyword export. It holds seven parts:
- The business definition, with the facts that change page decisions.
- The URL inventory, with a keep, update, merge, redirect, retire, or review action for every existing page.
- The Framenet entity map, with a central entity and a defined home for each supporting entity.
- The site structure, with parents, navigation, and internal-link relationships.
- The Page Map, with one job per candidate page.
- The variation rules for locations, categories, filters, and attributes.
- The open-question log, with an owner and an action for each unknown.
Each team then works from a different slice of the same source. The owner sees which services, categories, and markets the site will represent, and which proof they still need to supply. The developer receives routes, redirects, parent relationships, and the filters that stay out of the index. The designer receives page types and navigation. The writer receives a page-specific context: central entity, page type, query role, related pages, required proof, allowed claims, and open questions.
This is where Day 7 fits. The 8 Rules Every Money Page Should Follow make one approved page clear, direct, and useful. Domain-Wide Analysis decides whether that page should exist at all, what entity it owns, what facts it may state, and which pages support it. A page follows the rules. The site follows the strategy.
The strategy also stays current. When the business adds a service line, drops a product family, or enters a new market, the map changes first and the pages follow. That is Framenet’s fourth rule applied to the whole domain.
When is the website strategy ready to build on?
Before anyone builds, restructures, or writes for the site, I want nine conditions to be true:
- The business model, offer, operational scope, and commercial goal are recorded.
- Existing URLs have a deliberate action rather than an accidental future.
- Competitor research has produced questions and decisions, not copied page structures.
- Framenet has assigned one central entity and supporting entities with defined homes.
- The site structure gives approved pages a parent relationship and a navigation path.
- The Page Map records a distinct job, URL decision, and verification state for each candidate page.
- Keyword research supports the page’s job without creating the job on its own.
- Geography, catalogue filters, product attributes, and other variations have been judged at the right structural level.
- Open questions, issues, and actions identify what must be confirmed before the site makes a claim.
When these conditions exist, every change to the website becomes easier to review. The owner has a scope, the writer has a boundary, the designer has a page role, the developer has a route and relationship, and the reviewer has a standard to check. Everybody is finally working on the same site rather than five parallel versions of it.
That is what “done” means for a Domain-Wide Analysis. The next step is the Page Map, where every candidate page must answer one more question: does it deserve its own URL, or does it belong inside a page that already owns the reader’s need?




