Prioritize what to write next

The Demand page tells you what to write or update next. It ranks every signal Topicary collects into one worklist, so you decide from one screen instead of five.

Demand sits directly under Dashboard in the sidebar and is available on every plan, including Free.

Open the worklist #

1. Go to Demand in the sidebar.

2. Choose Refresh worklist to re-run gap detection and the health scan.

The page description reads Scanned plus a relative time once a scan has finished. If a fresh scan cannot be queued, the current worklist still loads and a note appears under the button.

Read a row #

Each row carries three things:

• Priority: one number out of 10, on the left.

• Item: the title, followed inline by the detector's reason, dimmed and shortened. Use the chevron at the right edge of the column to expand the full reason.

• Metadata line, under the title: the action chip (Create, Update, Expand, Verify, Merge, Split, or Archive), the source label, an evidence link where one exists, and the age glyph.

What the priority number means #

The badge shows a whole number out of 10. Higher means more urgent. Its tint carries urgency and nothing else: red at 7 and above, amber at 4 to 6, grey below 4. Action chips are grey on this page, so color always means urgency here, never a category.

The score blends three inputs:

Hover the badge to see Priority n/10. When content debt contributed, the tooltip names the debt score too. Read Content debt score for how that number is built.

The score has no time term. Age is tracked separately.

Spot overdue items #

The age glyph on the metadata line shows how long an item has waited, as {days}d. When an item passes the threshold for its priority, the glyph escalates and the item moves into Do now.

At 1 times the threshold the glyph is a muted clock, at 2 times a warning triangle, at 3 times a red one. Its tooltip states the fact rather than a demand: Raised 21 days ago — past the 7-day mark for this priority.

Work Do now, then Suggested #

The live list splits into two sections:

• Do now: items you picked up, plus items that are overdue for their priority.

• Suggested: the ranked backlog you pull from.

Both sections hold under either sort, because they describe commitment rather than score. Every item lands in exactly one place, by this order:

Section headers count items, not visible rows, so expanding a group never changes the number.

Switch views #

Six pills sit above the list. They are single-select: choosing one swaps what the table shows.

Counts on All, To create, and To improve cover the live worklist only, so Later and Snoozed items are never counted twice. Later and Snoozed count the whole project, so narrowing the list does not shrink them. Done carries no count because it is fetched only when you open it.

To create and To improve do not combine with Later, Snoozed, or Done. Those three are flat lists.

Narrow the list #

The toolbar under the pills holds three controls, hidden only on the Done view:

• Search: matches the title or the reason.

• Sort: Priority or Newest.

• Effort: Any, Quick wins, or Deep work.

Effort comes from the action, not from an estimate in minutes. Verify and Archive are quick wins, Create and Expand are deep work, and Update, Merge, and Split sit between the two, so they appear only under Any.

Search and Effort combine with every view except Done.

Act on an item #

Most rows promote one labelled button. Which one depends on the item:

The button you press holds its spinner through the editor load, which takes a second or two. A row with nothing to open, such as a code change whose topic was deleted, shows no primary button. Snooze, dismiss, and the ⋯ menu still work.

Opening a topic from Demand carries the context with you: the editor shows a banner naming the task in plain language, and the relevant panel opens. Hide on that banner only clears it from the screen. The item stays on the worklist until you resolve it.

For a code change, the staged edits wait in the editor's Suggestions panel. See Track changes and suggestions.

Snooze an item #

Snooze defers an item without deciding against it.

1. Choose Snooze on the row. One click defers the item until its evidence changes.

2. To set a timer instead, open the chevron next to Snooze and choose 1 day, 1 week, or 1 month.

Every snooze menu ends with the same sentence: Moves to Snoozed — returns when its evidence changes. A timer snooze also wakes early when the evidence moves, whichever comes first.

Evidence means the action, the detector signal, the signal count, and the priority. The title and the reason are deliberately excluded, so re-worded detector text does not wake everything on the next scan.

To bring an item back yourself, open the Snoozed view and choose Wake.

Move an item to Do now #

1. Open the ⋯ menu on the row.

2. Choose Move to Do now.

The row moves into the Do now section and stays in the list. To release it, choose Move back to Suggested in the same menu.

Mark an item done #

1. Open the ⋯ menu on the row.

2. Choose Mark done on a health-scan recommendation, or Mark fixed on a docs test.

Reader gaps, quality checks, and code changes have no done entry. Resolve them by acting on the topic, or dismiss them.

Dismiss an item, and undo #

Choose the ✕ on the row to dismiss an item you do not intend to act on.

An inline bar appears above the list for 8 seconds with an Undo button. Undo restores a dismissed row. After an Archive the bar reads Archived <title> — it is no longer published or scanned, and undo un-archives the topic itself, not just the row.

Work a group of similar items #

When 3 or more items in the same section share a detector, they collapse into one group row with a plain-language summary, such as 12 topics aren't in any map, and a chip reading {n} items.

• Use the chevron to expand the group. Members render as normal rows, indented and tinted.

• An expanded group offers Snooze all and Dismiss all. Dismiss all works through the members in order and stops at the first failure, leaving the rest in place.

• Grouping applies only when you sort by Priority with no active search.

• Groups always start collapsed, including after a page reload.

See the reader queries behind a gap #

A reader-gap row shows a quiet link on its metadata line labelled with the count, for example 11 reader queries. Choose it to open the stored evidence: a volume tile, the individual queries, and a Browse reader questions button that opens the AI queries tab.

Review what you finished #

The Done view is a read-only history, 25 rows at a time, with columns Outcome, Action, Source, Item, and Closed. An outcome is Completed, Created, or Dismissed. A ~ before a closing time means the source stored no exact close time, so the time is approximate.

Where demand comes from #

Each item carries a muted source label naming the signal that produced it:

See also #

• Triage content gaps: the reader-gap signal feeding "To create"

• Automated lifecycle recommendations: the health scan feeding "To improve"

• Content debt score: how debt weights an item's priority

• Content health tab: the same authoring-side signals, grouped by type

• Track changes and suggestions: reviewing the edits a code change stages for you