Sprint Decisions

The decisions in force and the shape they set for the build. Nothing here is signed schema.

Scoring has two tiers

Tier one is universal brand health, always reported, the same five dimensions for every client regardless of industry: Content Strength, Narrative Ownership, Distribution Power, Community Strength, Monetization Infrastructure. DBM reads 33 on it.

Tier two is the vertical rubric, developed per category to score what matters in that industry. DBM's construction five reads 52.5.

Both numbers are correct and they answer different questions. The structure for holding both was specified on 2026-07-06 as a root definition, a dimension registry, and vertical rubrics selecting from it. Building that registry (SB-04) is the next step and it is already designed.

Canonical read for tier one: projects/dbm-global/deliverables/dbm-brand-read-v2/.

How a dashboard element is declared

An element names four things before it is drawn: where the intelligence comes from, how it renders, who reads it, and what the reader can do about it. A dashboard template selects and orders those declarations.

Two supporting rules:

1. A channel can subscribe to other channels. That is how research, scoring and negative-space mapping compose into one element.
2. Every declaration states what it attenuates as two numbers, so the reduction is measurable rather than asserted.

This vocabulary is a candidate. The team settles the working language together. KG's four questions are the plain-English version already in use: what requires a decision, what changed, what is at risk and where is growth available, what is being done.

The shipping set

Kept small so everything in it works end to end.

Readers	One, Thomas at DBM Global, real and observed. Others join when there is a named person
Action types	Two: arm a rule, open the evidence. Three more join as their destinations are built
Channels	Twelve, each with an honest status: live, live for one client, local only, or to build
Forms	Twenty, eleven of them already built and rendering reports
Intake steps	Three: extract, declare, adjudicate. The user story generates from the declaration

Where composition happens

Composing an element means writing eight fields into a file and opening a pull request. route-gate.py validates it, with thirteen checks passing, and it enters as the repository's first pull request so the check exists before the first declaration does.

That is enough until the volume justifies a composer interface, so the studio interface waits.

Design system

Three artifacts replace the hand-skin pass: a token file and a component directory owned by design, and a schema co-owned by design and engine. The finish line is a fourth report that generates, renders, and needs only a review.

Module identifiers are assigned at the pull request that builds each one, taking the next free number.

Working agreements

• Severity always carries a score out of ten, everywhere priorities stack.
• Anything that moves carries a motion spec before it ships.
• Every figure opens its evidence.
• Everything resolves to an action.
• Instrument readout, never theater.
• Cream field, editorial register.
• The reskin handoff format is the standing pattern: token block, fonts, component styles, where the copy lives, brand rules to keep.

What sets direction next

1. The repository name. Recommendation: platform.
2. Ranking the register.
3. Building the dimension registry.
4. The Overall Health roll-up method.
5. One band ladder.
6. Theme.
7. Founder OS shape: self-serve, internal or both; one product or a shell holding several.

Related: feature-register-INDEX · 2026-07-28-walked-chain-dbm · sprint-plan-2026-07-27