What it is

A short grid that names each growth engine and says whether it is live on this page, live on its own page, or still waiting on something outside this Admin. It is a map, not a dashboard: it holds no figures of its own and it changes only when the code behind an engine changes. Analytics holds measured facts — what happened; this page holds interpretation, opportunities, recommendations, action queues and impact learning — what to do next.

Why it helps Gone Viral

Gone Viral's growth tooling grew in steps. This card stops you hunting through pages or assuming a capability exists when it has not been connected.

Example

You want to know whether Meta advertising insight is available before you plan a campaign. The card shows 'Meta Ads Opportunity Engine — Live on this page', so you scroll down rather than opening Meta's own dashboard.

How to use / set up

Read it once when you land, then jump to the section you need with the 'On this page' links above. For measured reporting — website, search, store and ads figures — open Analytics instead. No setup is needed and nothing here can be clicked to change a provider.

What it is

Astra is the Admin's reasoning assistant. It answers in text using a fixed set of read tools over data this Admin has already measured, and it can only propose actions — every action goes through the Action Registry and the Agent Gateway, which re-check permission and evidence server-side. Four profiles (Normal, Operator, Creative, Red Team) change only the reasoning style, never the tools or permissions. Listen mode uses the browser's own speech recognition and speech synthesis; Astra Neural Voice is a real spoken conversation that routes every request back through the same Astra core and requires the voice key to be present.

Why it helps Gone Viral

It turns a wall of panels into a conversation: 'what should I do about installs this week?' becomes a structured answer that separates what is measured, what is stale or unknown, and what is inference — with the Admin's evidence bands rather than invented confidence percentages.

Example

You ask in Operator profile whether to increase Meta budget. Astra reads stored Meta insights and freshness, reports that there are zero campaigns and therefore Insufficient evidence, and proposes nothing — instead of pretending a budget change would help.

How to use / set up

Type a question and choose a profile; the conversation is saved so context carries across visits, and 'Start a new conversation' begins a clean thread while keeping the full record. Use Listen mode if you prefer to speak; use Neural Voice only when its status shows ready. Astra never receives a provider token, never spends money, and never executes an action without a person approving it in the Action Registry. If the model gateway is not reachable the panel says so plainly and keeps your question.

What it is

A read-only table of every measured source — search, answer engines, Bing, Meta, app stores, subscriptions and more — with the age of its last real measurement and the evidence band (Strong, Moderate, Weak, Insufficient) that band gives to any opportunity built on it. Sources that were never measured say Never measured; blocked sources say Blocked.

Why it helps Gone Viral

A recommendation built on three-week-old data is not the same as one built on this morning's. Freshness is what stops Astra and the opportunity engines presenting old evidence as current, and it is what forces a human approval when evidence is Weak.

Example

The AEO section suggests a page change, but Evidence freshness shows Search Console was last measured three weeks ago. You treat the suggestion as Moderate at best and re-measure before acting on it.

How to use / set up

Scan the 'attention first' rows; they are the sources most likely to be dragging confidence down. Nothing here triggers a refresh — go to Measurement refresh for that. A source that has never been measured says so plainly rather than showing a comfortable-looking age.

What it is

A coordinator that turns the freshness table into a plan. Each row is classified: safe to refresh from here, deliberately manual because it costs money (DataForSEO-backed competitor and trend data), not configured, or blocked outside this Admin. Safe refreshes run only when you click them and record a new stored check.

Why it helps Gone Viral

Without it you would either forget to refresh or accidentally spend on a paid provider. The plan makes the difference between 'stale' and 'unavailable' explicit and keeps billable calls under human control.

Example

Before a weekly review you open the plan, run the safe refreshes it offers (Search Console, Bing search analytics and Meta read insights) and skip the manual DataForSEO rows, which cost money.

How to use / set up

Click a safe refresh to re-measure that one source. Manual and billable rows show what they would cost in plain words and are never triggered automatically — not by Astra, not by the scheduler. Blocked rows explain the external step required and offer no bypass.

What it is

A ranking engine that reads the findings already produced by the search, funnel, ads and market engines on this page — plus the measured reporting on Analytics — and sorts them into a single priority list. Each row keeps a stable action ID so the same job is recognised again tomorrow, and each row says which section it came from.

Why it helps Gone Viral

Gone Viral has a lot of measurement surfaces and only one owner. This turns twenty dashboards into a short list of what to do next, in the order most likely to grow installs and Viral+ subscriptions.

Example

You open Growth Intelligence on a Monday, see 'Improve the Reel Simulator landing page — high impressions, low clicks' at the top, click through to the section that produced it, fix the page title, then track the result in impact tracking.

How to use / set up

Read from the top down, use the jump link on a row to open the section that produced it (measured sections open on Analytics), and use Export CSV/JSON to keep a copy of the current list. Nothing external is required — it is built from data already loaded in Admin. If evidence is missing it says Insufficient data rather than inventing a suggestion.

What it is

A summariser that turns the ranked actions into plain narrative: what changed, what it means and what to do. It is deterministic, meaning the same inputs always produce the same wording, so you can compare week to week.

Why it helps Gone Viral

Useful when you want the 'why' rather than the numbers — handy for briefing a contractor, a marketer or an investor without handing over the whole dashboard.

Example

Before paying for a week of ads you read the advisor summary, see that organic search demand for 'reel simulator' is rising while your click-through rate is falling, and decide to fix the page first instead of buying traffic.

How to use / set up

Read the summary after the ranked list. No controls are required. If an optional external language provider has not been configured, this section states Unconfigured and keeps showing the deterministic summary — it never invents commentary.

What it is

A ranking layer that merges the opportunities already produced by the search, AEO, Meta, app-store and subscription engines into one list with a coverage summary and explicit gaps. It reads stored results only; it does not re-run any engine and does not contact a provider.

Why it helps Gone Viral

Growth levers interact — a landing page that answers a common question can lift organic search, answer-engine visibility and paid conversion at once. Seeing every channel in one ranked list makes that trade-off visible instead of siloed.

Example

The list ranks 'Add a clear Reel Simulator answer block to the home page' above a paid idea because it is supported by both Bing search evidence and AEO evidence, while the paid idea rests on zero measured campaigns.

How to use / set up

Read the coverage summary first so you know which channels are actually feeding the list, then work down the ranked rows. Gaps are listed underneath — a channel missing from coverage means its opportunities simply are not there, not that they scored zero.

What it is

An analysis layer that looks for repeatable patterns in your own search data — pages ranking just off the first page, queries with demand but weak click-through — and states the fix for each.

Why it helps Gone Viral

Search Console tells you what happened; this tells you what to change, which is the part that actually moves rankings.

Example

It flags a page sitting at position 11 with solid impressions; a tighter title and a clearer heading is often enough to move it onto page one and pick up the clicks.

How to use / set up

Work top-down and treat each row as one job. Already configured through the existing read-only Search Console access. Where there is not enough search volume to be confident it says Insufficient data instead of guessing.

What it is

An analyser that reads the site's real pages plus measured search queries and scores how well each page answers the questions people actually ask — the signal answer engines and AI assistants use. Each run is stored as a snapshot with the sources it could and could not use, so results are comparable over time.

Why it helps Gone Viral

More and more discovery happens through answer engines rather than ten blue links. Pages that answer directly, with clear structure, get quoted; pages that do not are invisible. This engine points at the exact pages and questions to fix.

Example

A run finds that 'how does the reel simulator work' has real Bing impressions but no page answers it in the first paragraph. The engine ranks that as high priority with the evidence attached, and Astra can turn it into a content draft for approval.

How to use / set up

Click Analyse to produce a new snapshot; the previous ones remain in history. Each saved run records which providers supplied evidence and which were unavailable at the time, so a run never implies a source it did not read. Nothing is written to the website from here; page changes still go through the Action Registry and, where relevant, a GitHub draft pull request.

What it is

Trend direction for topics related to short-form video and virality, from the same cached external provider snapshot. It shows movement over time rather than a single point in time.

Why it helps Gone Viral

Publishing while a topic is climbing is far cheaper than competing after it peaks.

Example

Interest in a new platform feature starts climbing, so you publish a Gone Viral explainer that week and pick up early search traffic before rivals notice.

How to use / set up

Look at direction first, size second. Refresh is manual and paid; opening the page never triggers a paid call. Unconfigured provider access is stated plainly.

What it is

A cached competitor view built from an external paid search-intelligence provider. Data is stored durably and only ever refreshed when you press the refresh control, so no cost is incurred by simply opening this page.

Why it helps Gone Viral

Shows where rivals already win the search terms Gone Viral wants, so you can pick fights you can realistically win.

Example

A competitor owns a broad head term, but you spot a specific phrase like 'reel virality score' they ignore, and you build the page for it.

How to use / set up

Read the cached snapshot, and only press refresh when you genuinely want fresh paid data. Provider access must be configured externally; until it is, the panel states Unconfigured and shows nothing invented.

What it is

A country-level view built from your own first-party Google search data, ranking markets by measured demand and current performance. It uses Google's own data only — no paid provider is involved.

Why it helps Gone Viral

Shows where Gone Viral already has interest it is not serving, which is usually cheaper growth than fighting harder in a saturated market.

Example

You find strong impressions from a country with poor click-through, so you adjust wording and examples for that market and capture demand you already have.

How to use / set up

Look for high demand paired with weak performance — that gap is the opportunity. Already configured through the existing read-only search access. Countries with too little volume to judge are reported as Insufficient data.

What it is

An analysis layer over your real ads reporting that highlights waste and under-served demand. It is strictly advisory and cannot alter any campaign.

Why it helps Gone Viral

Small budgets punish waste heavily; catching a poor keyword early keeps money on the ones that convert.

Example

It flags a keyword taking spend with no conversions, you pause it in Google Ads, and the same budget goes to the term that produces trials.

How to use / set up

Read the recommendations and apply them yourself in Google Ads. This section stays silent rather than guessing while Google Ads reporting is Unconfigured.

What it is

A read-only analyser over Meta Ads insight rows (campaign, ad set and ad level). It produces the same recommendation for the same inputs every time, with a confidence derived from impressions and clicks, and stores each analysis as a snapshot. The Meta write gate is deliberately disabled, so no campaign, budget or status change can be made from this Admin at present.

Why it helps Gone Viral

When paid campaigns run, this shows which ones deserve budget and which are burning it, using Meta's own numbers rather than opinion. Until campaigns exist it truthfully reports zero rows so nobody plans around invented performance.

Example

After a first campaign runs for a week, the engine flags one ad set with strong click-through and low cost per install and another with spend but no installs, and recommends shifting budget — as a proposal for a human to approve, not an automatic change.

How to use / set up

Click Analyse to fetch a fresh read-only snapshot. Zero rows means no campaigns have run in the window — a valid zero, not a fault. Any Meta change Astra proposes appears in the Action Registry as unavailable while the write gate is off; turning that gate on is a separate, explicit owner decision and even then new campaigns are created paused and launches need a second approval.

What it is

A return-on-investment view that combines spend with measured results such as installs and Viral+ conversions, and states plainly when there is not enough evidence to draw a conclusion.

Why it helps Gone Viral

Answers the question that matters most for a self-funded product: is this spend paying for itself?

Example

Spend rises but paid conversions do not, so you cut the campaign for a fortnight and put the money into the content that already ranks.

How to use / set up

Read the summary and treat Insufficient data as a genuine 'not yet knowable', not a bug. It becomes fully populated once real ads spend data is configured and available.

What it is

One consolidated queue of search-related actions with stable IDs, so a job you saw last week is recognised as the same job today rather than reappearing as a new suggestion.

Why it helps Gone Viral

Stops the same fix being redone or forgotten, and makes it possible to measure whether the work actually helped.

Example

You complete three queued title rewrites, mark them done, and later confirm in Action Impact Tracking whether clicks to those pages improved.

How to use / set up

Work the queue in order and mark items as you go. Already configured. An empty queue means no qualifying evidence right now — a valid zero, not a failure.

What it is

The earlier impact records, kept for reference. New tracking happens in Action Impact Tracking; this view is deliberately read-only so history cannot be rewritten.

Why it helps Gone Viral

Gives you a long memory: you can see which kinds of change have worked for Gone Viral before instead of relying on recollection.

Example

Before rewriting another landing page you check history, see previous title rewrites produced steady click gains, and prioritise more of them.

How to use / set up

Browse only — there is nothing to configure and nothing to edit. Comparisons are association only: they show what changed after an action, not proof the action caused it.

What it is

The complete, fixed catalogue of actions the Admin can perform on Astra's behalf — content drafts, IndexNow submissions, GitHub branch/file/draft-PR work, Meta proposals and so on — each with its risk, the permission level it needs and whether it is available right now. The autonomy policy sets how much may run without a person; Weak evidence always forces approval and Insufficient evidence is blocked. Approvals are queued here, decisions are audited, and rollback data is kept where an action supports it.

Why it helps Gone Viral

It is the safety boundary. Because there is no free-form execution path, you can read this one panel and know exactly what Astra could do, what it cannot, and what is waiting on you.

Example

Astra proposes a GitHub draft pull request with a page copy change. It appears here as a pending approval with its evidence band; you approve it, the gateway creates the astra/ branch, writes the file and opens the draft PR — and main is never touched.

How to use / set up

Set the autonomy level and any category overrides deliberately; the emergency stop halts every action at once. Review pending approvals with the evidence shown, then approve or decline — every decision is recorded. Unavailable actions say why (for example the Meta write gate is off or a credential is missing) and never pretend to be ready.

What it is

Named run definitions (currently a Daily Growth Review and a Weekly Full-Business Growth Review) that the production scheduler executes when they fall due. A GitHub Actions workflow calls the Admin's scheduler endpoint hourly with a secret bearer token; the server decides what is due from each definition's cadence, runs it once (idempotently) and records a report. Scheduled runs read stored evidence only — they never call a provider, never spend, and anything needing action is queued for approval like any other proposal.

Why it helps Gone Viral

It keeps the opportunity picture current without you remembering to press buttons, while keeping every consequential step behind human approval. The proof is stored: the readiness state only shows Active because completed external-scheduler runs exist in the database.

Example

On Monday morning the weekly definition is due. The 07:17 hourly trigger runs it, the report shows seven opportunities found, zero actions executed and zero approvals queued, and last-run timestamps update. You open Astra and ask what changed since last week.

How to use / set up

Enable or disable a definition here; controls are available only while the scheduler token is configured. Use Run now for an immediate manual run across all scopes if you want a report straight away. The status line reads Not configured, Configured — not externally tested, or Active — GitHub scheduler connected, and it is derived from stored run evidence rather than from the presence of a secret.