For fifteen years, the keyword was the planning unit of SEO. Pick a phrase, build a page, repeat. That model is breaking down because the engines stopped thinking in phrases and started thinking in things.
Modern search engines and AI answer systems represent the world as a graph of entities. A person, a company, a product, a concept. Each entity has properties and relationships, and the engine matches a query to entities before it matches it to documents.
A page that ranks well is no longer a page that contains the right words. It is a page that is clearly about a specific entity, supports that claim with structured data, and connects to other entities the engine already knows.
Keywords are useful for measuring demand and for picking the language of a page. They are no longer useful as the unit of content planning. Two pages targeting different keywords for the same underlying entity will cannibalize each other, because the engine sees them as duplicates.
The new planning unit is the entity. One entity, one page, many keyword variations served by the same document.
Start with a list of the entities your work is about. For me, that includes a person (myself), an institution (IBA Karachi), an employer (10Pearls Pakistan), a set of methods (marketing automation, technical SEO, AI workflows), and a set of tools (n8n, Claude, Ahrefs).
Each entity gets one canonical page. Every other page either supports that page or describes a different entity. There is no third option.
Wikidata is the public scaffolding that search engines and AI systems use to resolve entities. Being on Wikidata is the strongest possible signal that your entity is real and disambiguated, but the bigger lesson is the shape of the data itself.
A Wikidata entry has a label, a description, aliases, types, and properties. Any page on your own site should be readable as the same kind of object. If it is not, the engine has to guess what your page is about, and guesses are not citations.
In a keyword world, internal links are about passing authority to money pages. In an entity world, they are about declaring relationships. A link from a project case study to the method page that describes how it was built is a sentence in the graph, not a vote.
Anchor text matters again, but for a different reason. The anchor is the engine learning what your linked page is about. Vague anchors waste a sentence.
List every entity on your site. Note which ones have one canonical page and which ones are scattered across several. Pick the worst offender, consolidate it into one page, and redirect the rest. Do this once a quarter and you will outpace any competitor still building keyword pages.
The /technical-seo and /about pages on this site are the version of that exercise I keep running on my own work.