The default content strategy is publish-and-forget. You write a piece, ship it Tuesday, watch the bell-curve open-rate over five days, then move on. Three months later the piece has aged out of every algorithmic surface and nothing on your site links to it from anywhere else. You paid full freight to produce it and you are renting the result. The brief that produced this work treats content as a chronology. Chronology is the wrong frame.
This essay is the playbook for the alternative. Content as a topical graph. Each piece is a node engineered for retrieval, surrounded by the other nodes that make it findable, and audited against a topical map you maintain on purpose. The model is the wiki, not the blog. The reference is Standard Model of Effective Media, where this piece is the operational chapter on the topical-authority component.
Why the feed decays and the graph compounds
A content feed has one structural property: time decay. The newest piece is the visible piece. The piece you wrote six months ago might be the best thing on the property and it is invisible because the feed gives it no surface. Search engines, recommendation engines, and human readers all behave the same way: they grade the latest item, judge the property by it, and bounce.
A content graph has a different structural property. Topical density. Pieces are positioned by what they cover, not when they shipped. New pieces inherit traffic from older pieces through internal links, shared concept tags, and breadcrumb hierarchies. Older pieces get re-surfaced when new pieces link back to them. The ten-month-old explainer on a foundational concept is the foot traffic generator for the new tactical guide that depends on it. The Karay-audited wiki entry on Subject X is what gives the field note on Subject X-prime any retrieval surface at all.
Andy's own corpus is the working example. Eight wiki entries, each with a position in a five-cluster topology, each with internal links to two or three other entries, each with case-study and field-note links pointing back. New work lands and immediately inherits the link equity of everything that came before. A field note from week 30 lights up a wiki entry from week 4. The compounding is structural, not promotional.
Concept-node mapping · the discipline that produces the graph
Concept-node mapping is the discipline. The deliverable is a map. The map is a list of canonical concepts your audience cares about, organized into clusters, with each concept labeled by its current state: covered, scheduled, gap. Before any piece gets briefed, you locate it on the map. Before any piece ships, you confirm it has at least two inbound links from existing nodes and at least one outbound link to an adjacent node.
The rules are mechanical.
- Each piece covers exactly one canonical concept. If the brief covers two, it is two pieces. The single-concept rule is what makes the node retrievable; it is also what makes the writing tighter, because you stop forcing tangents into the body to pad word count.
- Each piece links to its parent concept (one level up the topical hierarchy) and at least one sibling (a peer-level concept that shares a parent). This is what builds the graph topology. Parent and sibling links are the wiring; without them, the piece is an island.
- Each concept has a canonical entry. New tactical or applied pieces that touch the concept link to the canonical entry, never re-explain it. The canonical entry compounds traffic; the tactical pieces compound depth. You do not have ten pieces fighting for the same head term, you have one canonical and ten satellites.
- The map is audited monthly. Concepts that have moved up in importance get prioritized for canonical-entry depth. Concepts that have moved down get demoted to satellite status. The map is alive; it is not a one-time spreadsheet.
This is what the cartographer reference points at. Karay is the institutional name for the audit role. The cartographer is the operator who owns the map, runs the audit, and ships the canonical entries. Without this role, the property reverts to chronology by default.
Canonical vs satellite · how the two shapes compose
The graph is a two-tier structure: canonical entries and satellites. Each tier has a different job and a different cadence. Confusing them produces a property that does neither well.
Canonical entries are long, thorough, evergreen. The Wiki gold-tier essay is the archetype. A canonical entry takes the foundational concept and gives the reader the complete model: definition, mechanics, how it composes with adjacent concepts, what it costs to apply, what it returns. The Standard Model essay is one. People · Product · Process is another. These pieces ship rarely (one per concept, refreshed annually), they take serious effort, and they hold the search position for the head term. They are the load-bearing nodes. The graph hangs from them.
Satellites are short, tactical, current. Field notes are the archetype. A satellite picks one operational moment, names it, links it back to one or two canonical entries, and ships in a single sitting. Satellites do the daily attention work. They surface in the feed, draw the visit, and route the visitor into the canonical entry through the internal link. The conversion is satellite-to-canonical-to-trust. The satellite is the door, the canonical is the room.
The discipline is to produce both shapes, in the right ratio, with linked bidirectional. Most properties produce only satellites and wonder why nothing ranks for the head terms. A few produce only canonical and wonder why nothing brings daily attention. The graph wants both, and it wants them aware of each other.
The monthly audit loop
The audit is the move that separates a graph from a pile. A graph without audit reverts to a pile within ninety days, because the natural failure mode of any content operation is to ship the piece in front of you and skip the maintenance.
The audit loop has four moves, run on a fixed monthly cadence.
- Map review. Read down the concept list. Mark each row covered, scheduled, or gap. Note any concept that has changed in importance since the last review. Concepts gain importance when the audience starts asking about them in the chat surface, on calls, or in the inbound queue. Concepts lose importance when nobody asks about them for two cycles.
- Orphan check. Run a search on the property for any piece with fewer than two inbound links. These are orphans. Either link them in or retire them. Orphan pieces drag the average; they hurt more than they help.
- Link audit. Spot-check ten random pieces. Confirm each links to a parent and at least one sibling. Confirm the canonical entry for the concept exists. If it does not, schedule it.
- Gap prioritization. Of the gaps marked in step one, which two would unlock the most downstream traffic if filled? Those two get scheduled for the next month. Discipline is to schedule fewer than you can ship; the slack is what makes the graph improvements actually land.
The audit takes ninety minutes a month for a corpus under fifty pieces. It scales sub-linearly because the map is the bottleneck, not the corpus. A property of two hundred pieces still has a map of forty concepts. You audit the map.
What changes when content becomes a graph
Three things change in the numbers when the property runs as a graph. Each one has a fingerprint you can confirm in your analytics.
Older pieces stop dying. The standard chronological property has a half-life of about thirty days per piece. The graph property has no clean half-life because pieces get re-surfaced every time a new satellite links back. The canonical entries become permanent traffic floors. Andy's own corpus shows this: the People · Product · Process entry pulls steady weekly traffic from satellites shipped months later, not from its launch week.
New pieces inherit traffic from week one. A new satellite that links into three existing nodes shows up in retrieval surfaces immediately because the surrounding nodes already rank. The cold-start problem disappears. The piece does not have to earn its first reader; the graph hands the first reader over.
Briefs get shorter. When the canonical entry exists for the concept, the satellite does not have to re-explain it. One link replaces three paragraphs of background. The piece becomes more about the operational moment and less about the framework. Tighter writing, faster shipping, and the framework still gets credit because the link routes traffic to it.
The fourth thing that changes does not show up in a dashboard. The cartographer becomes a defensible role. A property with a map and an audit cycle is structurally harder to clone than a property that ships posts on a calendar. Anyone can copy the calendar. Replicating the map requires a year of audits.
Where to start
Three starting points, ordered by difficulty and impact.
Easiest, do today. Write down the ten concepts your audience asks about most. Mark each one covered, scheduled, or gap. The list is your first map. It is incomplete and the categories are arguable; ship it anyway. The next two starting points depend on having one.
Medium, this month. Pick the most important uncovered concept and write the canonical entry. Long-form, evergreen, thorough. Link it to two adjacent concepts on the map (parent and sibling). This entry becomes the load-bearing node for that part of the property. The next ten satellites you ship that touch this concept link back to it.
Hardest, this quarter. Retrofit internal links across your existing corpus. Walk every existing piece, identify which canonical concept it touches, and add the link to the canonical entry plus one sibling. This is the unglamorous work. It is also the work that turns an existing pile into a graph without rewriting anything. Property age becomes an asset instead of a liability.
Content strategy is concept-node mapping plus the audit loop that maintains it. The map is the artifact. The pieces are output of the map. For the broader theory of how this composes with attention, trust, narrative, and distribution, see Standard Model of Effective Media.