AI NativeMeasurement without action is bookkeeping. The action plan turns scan results into a prioritized work list: which gaps to close, in what order, with what type of action, and how to tell whether it worked.
How opportunities are scored
Every gap in a scan - a persona-stage-engine cell where you are absent or underrecommended - becomes an opportunity. Each opportunity gets a score that is the product of three factors:
- Value: how much this cell is worth to close. Combines demand signal for this stage, the persona's intent weight, and a business-weight configured in personas.
- Gap: how far below full recommendation you currently sit.
- Winnability: how achievable a lift is, based on competitor coverage strength and prompt content clarity.
The score and all three factors are shown on each opportunity item. The highest-scoring open opportunity sits at the top of the attack queue.
The attack queue
The attack queue contains open and in-progress opportunities ranked by score. Each item shows the cell, the score breakdown, a diagnosis, and three specific actions:
- Content: what page or piece to write, usually framed around the specific underperforming prompt.
- Distribution: which sources to get on, derived from the citation set for this query.
- Lever: the action type (schema, off-page, distribution, or amplify).
The "This week" view shows the top five open attack items from your highest-scoring product as a focused starting point.
The defend queue
The defend queue holds cells where you are currently recommended and where a shift could cost you that position. High value here means high exposure to loss, not high potential to gain. Defend items carry the same content, distribution, and lever actions; the interpretation is about maintaining rather than closing a gap.
Tracking progress
Each opportunity has a status: open, in progress, or done. You move it through these states from the run page or the action plan. When you mark something done, you can add a note describing what you shipped and which lever you pulled.
Marking done does not trigger a new scan. The platform waits for you to re-measure explicitly. When the re-measure finishes, the delta on the run page shows whether the action moved the number. This design makes the connection explicit: you shipped something, you re-measured, you saw the change.
Lever effectiveness
The action plan tracks which lever types have moved numbers from your completed, re-measured items. The average recommendation lift per lever type is shown as calibration for open item estimates. An unproven lever falls back to a conservative gap-based estimate, marked with a tilde.
Questions
How is the opportunity score calculated?
Score equals value times gap times winnability. All three factors and their product are shown on each item so the ranking is auditable.
What is the difference between attack and defend?
Attack items are cells where you are absent or underrecommended and the goal is to gain presence. Defend items are cells where you are recommended and the goal is to maintain that position.
Do I need to re-measure after marking an item done?
Marking done records what you shipped. Re-measuring shows whether AI picked it up. Without a re-measurement, the completed item shows a baseline and a prompt to re-scan.
Does the queue update when I run a new scan?
Yes. The queues always reflect the latest completed scan. New scans may add items, retire resolved items, or re-rank existing ones.
What does the projected impact estimate mean?
It is the estimated recommendation lift in percentage points if you ship the action for an open item. Where similar lever types have been completed and re-measured, the estimate is calibrated from those results. Unproven levers use a gap-based estimate marked with a tilde.
Can I see all opportunities across all products?
Yes. The action plan views at /app/plan/week, /app/queue/attack, and /app/queue/defend pull from all products in your workspace. Each item shows which product it belongs to.