Dataset documentation

First 100 days as a new Chief Compliance Officer: schema 1.0.0

How to read the files under /data/cco-100-day-plan. The JSON Schema is the contract the build enforces; the CSV tables are projections of the same approved records.

Not legal advice. The plan is original practitioner guidance, not a compliance assessment or an official checklist. Day bands are suggested planning phases, not deadlines. Firm types express relevance, not legal applicability. Session checkmarks are never part of this data.

Human page: /resources/first-100-days-chief-compliance-officer. Not yet released.

Files

Conventions

Dates: ISO YYYY-MM-DD. A checked date is the day a person opened the source; it is never a fetch or build time. Legal dates keep their stated meaning.
Amounts: Decimal text in the source currency, never floating point, never converted. An absent amount is null (empty in CSV), never 0.
Unknowns: Every fee, timeline, amount and fact carries a status. "Unknown: not verified" means nobody has checked; "Not published" means an official check found nothing published; "Not applicable" needs affirmative official evidence.
Ids: Lowercase slugs that never change once published. Display names may change; ids do not. Withdrawn records keep their id and address.
Sources: Exact official HTTPS URLs on an allowlisted host, with a locator and the supporting passage. A URL alone is not verification.
Review: Public files carry review status, date and a content hash; the reviewer is recorded privately, never published.
CSV: RFC 4180, UTF-8, CRLF, header row. Multi-valued fields are joined with " | " and bracket their source ids. A cell that a spreadsheet could read as a formula (leading =, +, -, @, |, %, tab) is prefixed with an apostrophe on export; pure numbers are not. The apostrophe is an export guard, not part of the value.
Caching: v1 files are mutable and served with a short public cache (five minutes) and revalidation; a correction replaces page, JSON and CSV in one build.
Reuse: Undecided while draft; no licence is implied. Underlying official documents keep their own rights.
Note: Tasks of kind guidance have empty source_ids and last_checked by design: they state no obligation. A task of kind requirement must carry a binding official source, locator and checked date or the build refuses it.
Note: The session_mark column exists only in a visitor's own current-view export, when they opt in; the static v1.csv never contains it.

Tasks table (v1.csv)

v1.csv

Columns of v1.csv, in order.
Column	Meaning
`id`	Stable task id; the page anchor is #<id>.
`phase_id`	As named.
`phase_label`	As named.
`day_start`	Start of the suggested planning band. Not a deadline.
`day_end`	End of the suggested planning band. Not a deadline.
`order`	Suggested order within the phase.
`title`	As named.
`explanation`	Plain explanation of the suggested work.
`output`	The concrete output, prepared in the firm's own systems.
`firm_types`	Relevance tags joined with " \| ". Relevance, not legal applicability.
`perspectives`	cco, founder, or both, joined with " \| ".
`topics`	Joined with " \| ".
`kind`	guidance (original practitioner suggestion) or requirement (sourced obligation).
`applicability`	Where applicability still needs confirmation.
`caveats`	Joined with " \| ".
`source_ids`	Empty for guidance.
`source_urls`	Same order as source_ids.
`last_checked`	Empty for guidance; the date a person checked the source for a requirement.
`status`	active, superseded or withdrawn.
`offer_path`	A contextual Taft page, if any.
`depends_on`	Suggested predecessors, joined with " \| ". Nothing is locked.
`review_status`	unreviewed, in-review or approved. No reviewer name is published.
`review_approved_at`	As named.
`revision_hash`	As named.
`session_mark`	Only in a visitor's own current-view export, and only when they chose to include it: Marked done or Not marked. Never in the static file; marking a task done does not establish compliance.