Skip to content

The Precedent handbook

Insights

The Insights page answers the questions a lead asks about the record itself: how much review debt exists, how long drafts sit, how often the team changes course, and which areas have no recorded decisions at all.

Use the page

Choose Insights in the sidebar. Start with the overview, then read the age and per-area breakdowns to identify what needs attention.

The governance report near the end of the page has its own area and period filter; the rest of the page always covers every area you can see.

Until there is a decision you can see, the page shows only Coverage by area and Types in use, which says its counts appear once decisions are recorded. The other sections appear with the first decision. If every decision has been archived, the governance report and review outcomes stay when a review of one of them recorded an outcome, because those reviews still count.

For overdue work, open Inbox. To find incomplete records, use Search and the area filters. Insights summarises the record; fixing it means reviewing, completing, or recording decisions in the app.

What it measures

Eight metrics, all computed from the record, the review system and the assistant call log, each except Types in use with a per-area breakdown at the bottom of the page.

Review debt. Open review items by reason (review date reached, awaiting approval, dependency changed, condition met) and by age. At Acme Robotics, three items over three months old meant three decisions nobody owned anymore; the fix was reassignment, not more process.

Time to decide. The median and 90th-percentile days from a draft’s creation to the day it was decided, plus the drafts waiting right now. A rising median usually means assistants are proposing faster than owners are reading.

Course changes. Decisions decided against supersessions and reversals, by month over the last year. A low number of changes can mean choices still hold, or that changes are not being recorded. When Acme Robotics eventually outgrows AR-041 (Postgres full-text stays as the search engine), the supersession lands on the record with its reasons, and this chart is where that shows up.

Review outcomes. Per area, the reviews with a recorded outcome, how many were reaffirmed and how many were reversed. Reaffirmation rate is the share reaffirmed; reversal rate is the share reversed without a replacement. Revisions, supersessions and archives count as reviews but in neither rate. Unscheduled reviews (a decision reaffirmed, superseded or reversed while no review was open) count; reviews resolved before outcomes were recorded are left out. At Acme Robotics, AR-041 (Postgres full-text stays as the search engine) was reaffirmed at each review so far, so the Catalogue and Infrastructure rows each count those reviews as reaffirmed: a high reaffirmation rate rather than an absence of data.

Orphaned decisions. Decided records with no links and no evidence. They say what was decided but not why, and nothing connects them to the decisions they constrain.

Coverage. Decisions per area, and areas with no decision activity in the last six months. A quiet area means one of two things: nothing was decided there, or decisions are being made without being recorded. Both are worth a conversation.

Where decisions start. Where decisions enter the record: the web editor, Slack, or an assistant. Every decision carries its channel from the moment it is created, so this needs no extra instrumentation.

Types in use. All seven Types, with the decisions you can see of each: drafts, decided, and for Exception, How-to and Fact the decided ones past their expiry date. For How-to and Fact it also shows how many of the decided ones an assistant opened with get_decision in the last 30 days, such as 12 of 20, each counted once however often it was opened. The other Types are not counted there: assistants receive rules, lessons and terms in full with the Area, so they rarely open them one at a time.

Governance report

The Governance report section answers an auditor’s questions for one area or all, over a period you choose. Pick an area and a date range and choose Apply; the period defaults to the last 90 days and can span up to two years, as for the audit export. The dates are UTC days, and the date pickers offer up to one day after today in UTC, so someone ahead of UTC can choose their own date. The filter is in the address, so a filtered view can be shared as a link.

Exceptions. Decided records whose Type makes the review date an expiry, such as Exception: how many stand now, how many expire within 30 days, how many have expired, and how many were reaffirmed with a new date in the period (one record reaffirmed twice counts twice). Every expired one is listed by ref with the date it stood through and its owner. At Acme Robotics, AR-118 (raw inspection images kept for 30 days instead of seven) appears there the day after its date unless someone reaffirms it with a new date.

Approvals. Per area, then for all areas: decisions decided in the period, how many were decided by the owner alone and how many with another person’s approval, requests for changes made in the period, and the median days from asking for approval to being decided. A decision tagged with two areas counts once in the all-areas row.

Reviews. Decided records past their review date with the review still open, with the oldest, the median and how many are past by a week, a month, a quarter or more. Expired exceptions are counted under Exceptions instead. Reviews with a recorded outcome in the period, by outcome, with the reaffirmation and reversal rates.

By Type and by Area. Decisions recorded and decided in the period per Type, listing every Type (a client portal lists only the Types its decisions carry), and per area the same with standing and expired exceptions, overdue reviews, reviews and reversal rate. A decision tagged with two areas counts in both rows.

Counts marked now are as of the day the report is built, in UTC. Period counts use the dates you chose, inclusive.

Each table has a CSV link, and Download PDF saves the whole section. Both are built from the same rows as the page for the same filter. The PDF names the workspace, the area, the period, when and by whom it was generated, and the id and SHA-256 fingerprint of the audit-export signing key, so it pairs with a signed audit export of the same period (Settings → Data). Every download is logged with the workspace’s exports and appears in the next audit export.

Visibility

Every count applies the same area visibility rules as the lists and search. A member who cannot open a restricted area does not see that area in the coverage, outcome or governance tables, cannot choose it in the governance filter, and decisions on it are absent from every total, share, and trend they see. The numbers differ between people by design: a workspace administrator sees the whole record, a member sees the record they can access.

This is applied inside the aggregation, not as a filter on a drill-down. The totals describe the decisions visible to the person viewing the page.

Reading it as a lead

A screenshot of the page fits a leadership deck: the tiles show the headline figures and the sections below explain them. Three patterns are worth acting on: review debt older than a month (decisions without effective owners), a mix of decision sources that is all web editor (the record is written after the fact, not at the moment of deciding), and quiet areas that ship changes anyway (an unrecorded decision stream).

Discard unsaved changes?

Your changes have not been saved. Stay to keep editing, or discard them and leave.