Action Center

The Action Center is the AnswerixAI page that turns your AI visibility data into a to-do list. It tracks the KPIs you set goals for, records signals when something meaningful changes, and groups those signals into actions with concrete tasks. It is for anyone who needs to decide what to work on next for a brand. Open it from Action Center in the sidebar (/dashboard/action-center).

What are the Action Center tabs?#

The AnswerixAI Action Center has four tabs: KPIs compares your key metrics with your goals, Actions lists what to do now, Signals shows what changed, and History shows what was done and what it moved. Each tab has its own address, so you can link to it.

The Signals tab shows a count of new, untriaged signals next to its name. Every tab shows data for the brand selected in the brand switcher.

What is the difference between a signal and an action?#

A signal is an observation: one meaningful change AnswerixAI detected in your data, such as a sharp visibility drop or a competitor passing you. A signal says what happened. It doesn't say what to do.

An action is a work package built from one or more signals. AnswerixAI groups related signals into a single action, so six pages without AI traffic become one "Capture AI traffic" action with six linked signals, not six separate items. Each action comes with a goal and a task list.

How are signals and actions generated, and how often?#

AnswerixAI generates Action Center signals and actions once a day, after your brand's full daily tracking run finishes. On-demand runs and single-prompt runs don't generate them. Each day AnswerixAI detects signals, builds actions from open signals, and measures closed actions, in that order:

1. Detect signals. AnswerixAI compares the last 7 days of tracking results with the 7 days before and records any condition that crossed a threshold.
2. Queue findings. Every action definition that matches open signals (New or Acknowledged) gets a waiting candidate, whether or not anything can be done about it tonight.
3. Build actions. The highest-priority candidates become actions, subject to the rules in How does AnswerixAI decide which actions to open? below.
4. Carry out automatic tasks. Tasks that a tool can do by itself, such as reading your AI visibility for the validation window, are carried out and recorded (see Which tasks does AnswerixAI carry out automatically?).
5. Measure closed actions. Each day, AnswerixAI checks completed actions whose measurement window has ended and records their result.

A brand's very first tracking run is an exception to "once a day": onboarding starts it by hand, and AnswerixAI runs one signal pass straight after it completes, so a new brand's Action Center is not empty until the next night. Only the first run gets this; later on-demand runs wait for the nightly pass.

Signals come from the same detections as the Daily Pulse, but they are recorded even if you turned Daily Pulse emails off. If a daily run has no fresh results, or an AI platform has a collection outage, AnswerixAI skips detection rather than treating the silence as a change.

Each nightly pass is logged against the tracking run it covered. If a pass fails partway, for example because a database statement timed out on a very large brand, the next morning's daily sweep re-runs it, so a lost night is recovered rather than lost for good.

Which signals does AnswerixAI detect?#

AnswerixAI detects 61 kinds of Action Center signal: the fourteen original detectors below, plus 47 detectors behind the action definition library, grouped into twelve families. Each signal has a type, an impact level and a source (AI Results, Google Analytics, Google Search Console or Site Audit).

The original detectors#

The definition library's detectors#

The other 47 detectors read compact summaries computed inside the database, one per family, so a large brand's nightly pass stays fast. Each family only runs when the brand has its data source, and a family whose read fails for one night leaves its open signals alone rather than marking them resolved. A prompt counts as tracked once it has results on 3 days of the window, or 5 answers in total, so a brand's first run already produces findings. Most findings about a group (prompts, pages, queries) become one signal that names its targets, such as Absent from 14 prompts; findings about one thing (a topic, a platform, a source, a competitor) are one signal each, capped at the 5 strongest per kind per night. A finding that slips past the cap keeps its open signal open but never opens a new one.

Conditions that persist, such as a visibility drop, stay as one signal and update each day. When the condition ends, AnswerixAI marks the signal Resolved automatically. One-time events, such as a first citation, never resolve on their own.

How do I work with signals?#

The Signals tab has five summary cards: Total Signals, Important Signals, New Signals, Resolved Signals and Avg. Signal Impact, each compared with the previous period of the same length.

The signals table has the columns Signal, Type, Impact, Source, Affected, Detected, Status and Actions. Click a signal to open its details: what changed, what it affects, the related KPIs, when it was first and last detected, and a View prompt link where a single prompt is involved.

How do I filter signals?#

You can filter Action Center signals by date range (7d, 30d or 90d, with 30d as the default), by type, by impact, status and source, and by text with Search signals….

• Date range: 7d, 30d (default) or 90d.
• Type: pills for All, Visibility, Citation, Mention, Traffic, Technical and Competitor.
• Impact, Status and Source dropdowns.
• Search signals… to match text.

Click Clear filters to reset them.

What do signal statuses mean?#

An Action Center signal has one of four statuses: New (not reviewed yet), Acknowledged (seen, and still feeding actions), Resolved (the condition ended) and Dismissed (not relevant, and stays dismissed even if the condition fires again).

To change a status, open the row's ⋯ menu or the signal details and choose Acknowledge, Mark resolved, Dismiss or Reopen. A resolved signal whose condition returns opens again as New.

How do I work with actions?#

AnswerixAI's Action Center has a library of 62 action definitions in five categories: 19 Growth, 10 Protect, 11 Recover, 13 Fix and 9 Compete. Each definition has a stable code (G01…C09), listens for one signal kind, and needs certain data before it can open. Four definitions are registered but switched off until the data they need exists: brand associations (G11) and brand-information consistency (F08) need attribute extraction, organic-page expansion (G13) needs page-level Search Console data, and broken citation targets (F07) need URL health checks. Each action is built from related signals, so you open the action, work through its tasks, and set its status as you go.

Every action title names its targets, such as Recover lost citations on 7 prompts or Close the visibility gap with 2 competitors; the count is snapshotted from the signals when the action opens, and an action raised with no countable targets uses its plain title instead.

An action's impact is the highest impact among its signals. While an action is open, new matching signals are linked to it as evidence, but its scope doesn't change. After you complete or dismiss an action, AnswerixAI waits 14 days before it opens a new action of the same kind, even if the condition is still firing.

Each action records the version of the definition that raised it, so an action opened before a rule was tuned is still read against the rule it was raised under. If AnswerixAI raises an action from a detection rule newer than your app version, the action still appears, titled from its rule name, with its tasks, evidence and measurement intact.

How does AnswerixAI decide which actions to open?#

Every night, each action definition that matches open signals gets a candidate: a record of the finding, covering the matched signals no action has claimed yet. A candidate then becomes an action only when all of these hold:

• No action of that kind is currently open (New, In Progress or On Hold).
• The brand has the data the definition needs (see the Needs column above). An action already open keeps going even if that source is disconnected later.
• The last action of that kind closed more than 14 days ago.
• The brand has not already been given 5 new actions today. An action opened this morning and dismissed by lunch still counts.

A finding that cannot be acted on tonight is not lost: it stays in the queue, and waiting raises its priority. Priority orders candidates by impact first, then by how many signals stand behind them (up to 10), then by how long they have waited (up to 14 days). The weights are chosen so that a high-impact finding is never outranked, and a low-impact finding that has waited two weeks outranks a medium one found tonight. A candidate whose signals resolve goes stale; if the condition returns, it comes back with a fresh clock.

How is an action's task list built?#

Every task comes from a shared task library of 62 primitives, from Analyze AI responses and Identify content gaps through Create a content brief, Publish the approved content and Send the approved outreach to the validation tasks that end every plan. Each definition lists the tasks it plans, and AnswerixAI plans them per brand, not per action kind:

• Data the brand has. Each task declares the data it needs, so a brand with Search Console connected is asked to inspect its queries, and a brand without it is not asked a question it cannot answer. Missing optional data removes a task, never the action.
• Optimise or create, never both. Where a plan branches, AnswerixAI optimises the page that exists when the action's prompts have a page mapped to them, and briefs and drafts a new page when none does.
• One task per primitive. Two parts of a plan that imply the same analysis become one task over the combined scope.
• Dependencies. A task that waits on another only lists dependencies that made it into the plan.
• Targets. Tasks name what they act on in their title, such as Analyze citations across 11 sources, using the same targets as the action.

Each task also records who carries it out and what it touches. Mode is one of system (a deterministic comparison AnswerixAI does), agent, human, or human or agent. Permission is read, write (changes your own content) or execute (acts outside AnswerixAI, such as publishing or sending outreach). Publishing is always a person's task, because no publishing integration exists yet, so the action is never blocked for want of one.

Which tasks does AnswerixAI carry out automatically?#

Every night, after the signal pass, AnswerixAI attempts the tasks that a tool implements, on actions that are still open, oldest first, up to 50 attempts across all brands. Two tools exist today, and both only read:

• Visibility window reads your AI Visibility over the 7-day validation window and the 7 days before it. It carries out Validate AI visibility.
• GA traffic window reads sessions on the action's pages over the same two windows, live from the brand's connected Google Analytics property. It carries out Inspect AI-referred visits and Validate AI-referred traffic when the brand has Google Analytics connected.

Four rules govern automatic execution. Nothing reaches outside AnswerixAI except through a tool. A task whose permission is write or execute, or whose tool would change something outside AnswerixAI, is held in Waiting Approval until a person approves it, and the approval is recorded on the task. Every attempt is recorded, with its inputs, tool, result or error, and never a model's reasoning. A tool that fails fails only that task, which is marked Failed for someone to do by hand; the action and its other tasks are untouched.

The action's History tab shows each automatic outcome as carried out automatically, the automatic attempt failed; it needs doing by hand, or waiting for your approval before it reaches outside AnswerixAI. Operators can run or dry-run a single task from the command line with php artisan answerix:run-action-task <taskId> [--dry].

What does the Actions tab show?#

The Action Center Actions tab shows five summary cards (Top Actions, At Risk, Opportunities, In Progress and Potential Impact) above a table of actions with their impact, status, assignee, due date and task progress. Click a card to filter the table to the actions it counts:

• Top Actions: open actions with high impact.
• At Risk: open Recover and Fix actions.
• Opportunities: open Growth actions.
• In Progress: actions with the status In Progress.
• Potential Impact: the highest impact across open actions.

The table has the columns Action, Impact, Status, Assignee, Due date, Updated and Actions, with each action's task progress. Filter by category (All, Growth, Protect, Recover, Fix, Compete), Impact, Status and Assignee, or use Search actions…. Sort orders the table by Priority (default), Impact, Newest, Due date or Recently updated. Priority means highest impact first, then the most linked signals, then the most recently updated.

How do I change an action's status, assignee and tasks?#

Click an Action Center action to open it, then change its Status, Assignee, Due date and tasks from the action itself. The action has four tabs: Overview (why the action exists, its goal and related KPIs), Tasks, Signals and History. From the action you can:

1. Set Status to New, In Progress, On Hold, Completed or Dismissed.
2. Pick an Assignee from your team.
3. Set a Due date.
4. Update each task to To Do, In Progress, Waiting Approval, Completed, Skipped or Failed. Skipping a task asks why it was skipped. Long task texts wrap in full rather than being cut off.
5. Rename or delete tasks, or click Add task to add your own (up to 200 characters).

Every change is recorded in the action's History tab, including the measured outcome once it is known. Skipped and failed tasks don't count toward the task total.

How do I set KPIs and goals?#

The Action Center KPIs tab tracks up to five metrics, AI Visibility, Citations, Mentions, Share of Voice and AI Referral Traffic, each with a goal you set for a weekly, monthly or quarterly period. Only admins and managers can change KPIs.

If no KPIs are set yet, click Configure KPIs. Otherwise click Edit KPIs or Add KPI. The Configure KPI Framework panel has two steps:

1. Choose KPI template: AI Search Visibility selects AI Visibility, Citations, Mentions and Share of Voice. Custom Framework lets you pick any KPIs. Content Performance and Brand Monitoring are marked as coming soon.
2. Set goals & targets: choose a Time period (Weekly, Monthly or Quarterly) and a target for each KPI. Targets must be positive, and percent targets can't exceed 100.

Click Save & Apply. Only admins and managers can change KPIs.

The KPI table shows KPI, Value, Change, Trend (7D), Status, Goal and vs Goal. Switch the date range between 7d, 30d (default) and 90d, and filter by category or status. When you view a window that differs from the goal's time period, count targets scale to the window; percent targets stay the same.

How is KPI status decided?#

AnswerixAI decides KPI status by comparing a KPI's value with its target: Goal Reached at 100% or more, Off Track under 40%, At Risk under 70% or when moving the wrong way, and On Track otherwise.

How does History measure results?#

The History tab lists actions that are In Progress, On Hold, Completed or Dismissed. Summary cards show Total Actions, Completed, In Progress, No Improvement and Dismissed. Filter by date (7d, 30d, 90d, All), Status, Type, Impact and Source, sort by Newest, Oldest or Impact, or search.

For each completed action, AnswerixAI compares the action's metrics over the 7 days before it was created with the 7 days after it was completed. The result appears once that after-window has ended:

Open an action in History to see Visibility Change, Citations Change, Prompts Improved and Time to Close. AnswerixAI reports what changed after the action closed, not that the action caused the change.

Which plans include the Action Center?#

The Action Center is available on every paid plan. Some signals and actions depend on other features. AnswerixAI resolves eight data sources per brand: tracking (always), competitors configured on the brand, completed Site Audit runs, an active Google Analytics connection, synced Google Search Console queries, prompt volumes on at least one prompt, AI traffic from Google Analytics or the tracking snippet in the last 30 days, and topics. A definition, a detector family or a task that needs a source the brand lacks is skipped for that brand; connecting the source later makes it eligible the next night.