Marketing

Marketing is where the platform turns what it already knows — what is selling, what is trending, who owns what, what just launched — into campaigns, audiences and outbound messages. The area covers three things: campaigns (a coordinated push around a title, a topic or a moment), audiences (the sets of customers a campaign speaks to), and outbound communications (the emails and notifications that actually go out).

Two ideas make this different from bolting a mailing tool onto an export. First, marketing here is built on the platform graph: the signals that drive an audience and the facts that fill a message are the same records the rest of the platform runs on. Second, marketing is rights-aware: putting content into marketing — an excerpt in an email, an AI-written blurb — is a governed content use, checked and tracked exactly like any other use of that content.

Built on the platform graph

A typical marketing stack lives beside the business: customer lists are exported into it, product copy is pasted into it, and both drift out of date the moment they arrive. The platform takes the opposite approach — marketing reads the graph directly, so an audience or a message is only ever as stale as the records behind it.

Each row is a live platform record, not a copy. That changes what an audience is. "Customers who bought a Go title in the last year but have no active subscription" is not a spreadsheet someone maintains — it is a query over orders, entitlements and product data, and it re-evaluates as those records change. The performance side works the same way: the platform already lands search and usage signals in an analytics store and periodically aggregates them into per-Resource metrics (recent sales performance, for example), and it maintains system labels like trending on its own schedule. A campaign brief that says "promote what is trending in machine learning" resolves against those platform-computed facts, not against a marketer's intuition.

Because the platform is event-driven, the freshness works without anyone polling. A new product entering a catalogue, a promotion going live, a label being applied — each is a signal Marketing can react to, the same way Distribution reacts to catalogue changes today.

Rights-aware marketing

Here is the idea that makes this area unusual: using content in marketing is using content. An excerpt quoted in an email, a chapter teaser on a landing page, an AI-written blurb summarising a book — each of these takes an author's work and puts it in front of people. On this platform that is a governed use, and the same machinery that governs every other use applies.

Concretely:

• AI-written copy is switched by the contract. When AI generates a blurb or a summary of a Work for a campaign, that generation is governed by the same AI_SUMMARISATION permission an author's contract grants or withholds — the switch described in AI and your content. An author who has said no to AI summarisation has said no everywhere, including in the marketing department.
• Attribution rules follow the content. If a contract requires attribution when content is displayed, an excerpt in an email carries the author's credit for the same reason an AI answer does — the rule is attached to the Work, not to the surface.
• Generation is an Activity. AI-produced product information is already modelled as platform work — generated blurbs are created by a recorded Activity, so what was generated, from which content, is traceable rather than folklore.

The payoff is symmetry. Authors do not need a separate conversation about "marketing uses" of their work: the contract they already hold answers the question, and the Content Lake's resolved rights are what any marketing feature checks before an excerpt or a generated blurb ships. Marketing gets speed — copy on demand, excerpts without a clearance email thread — precisely because the permission check is automatic.

The shape of the area

The working model has three layers, from most to least settled:

1. Inputs (settled) — the platform records in the table above. Marketing consumes them read-only; it owns none of them.
2. Audiences (direction settled, mechanics in design) — declarative segments over those inputs that re-evaluate as the graph changes. The search estates that would let audiences query customer and order data directly are planned but not yet built, so the first audience mechanics are part of the current design work.
3. Campaigns and outbound (in design) — composing a campaign from an audience, a promotion and content; scheduling it; sending messages; measuring the result. This is the layer with the most open questions.

If the area publishes events, they will follow the platform's naming grammar (platform.<service>.<entity>.v<major>.<verb>). The names below are illustrative, not yet claimed — they show the shape, not a commitment:

One boundary is already firm: Marketing reads performance signals but does not touch the money. The royalty model — what an author is owed when content is used — lives on contracts, and settlement belongs to the future Royalty service. A campaign that drives sales shows up in the same order and usage records as any other sale; it never gets its own side-channel accounting.

Related pages

• AI and your content — the AI_SUMMARISATION switch and attribution rules that govern AI-written marketing copy.
• Catalogues — where "what is on sale, at what price, in which destination" is defined.
• Royalties — the money model marketing reads around, never through.
• Events — the event-driven backbone Marketing will consume and, eventually, publish to.
• Content Lake — where the resolved rights behind every excerpt and blurb are held.