Crawl Foundry
Dashboard

Turn Home recommendations into verified work without inventing an activity feed

What to do now is a capped recommendation queue, not an assigned backlog. Home does not render a recent-activity feed. Verify each row in its owning feature, use the dedicated Activity page for recorded operations, and keep implementation and outcome measurement separate.

Recommendation, not task

The visible order is a heuristic over a few crawl and keyword signals, without effort, revenue, ownership, or release context.

Activity is elsewhere

The Home footer links to Activity and may show one keyword-event date. It does not render statuses, counters, failures, or history.

Verify the handoff

Follow each row to Site Audit, Position Tracking, Action Center, or the Keyword Database before assigning work.

Separate recommendations, ranking summaries, and recorded activity

Home uses three different ideas that should not be merged. What to do now derives a short recommendation list from selected crawl and keyword signals. What moved and Your rankings summarize two tracking contexts. Activity is only a compact footer destination on the current screen. None of these areas assigns a task, proves a cause, or records a business outcome.

Use the recommendation or ranking summary to choose an owner. Use the owning feature to verify the evidence. Use the dedicated Activity page to reconstruct supported operations. Keep the human decision, assignee, change, and success condition in the workflow that actually owns the work.

Read What to do now as a capped recommendation queue

The panel shows at most three rows under the current dashboard policy. The source queue is ordered: critical Site Audit findings, high Site Audit findings, the strongest tracked keyword loss, medium Site Audit findings, critical or high workspace opportunities, keywords missing own-rank evidence, and then low Site Audit findings. The display limit can exclude valid lower rows.

This order is a product heuristic, not a business-priority calculation. It does not incorporate implementation effort, affected template value, conversions, revenue, legal review, planned releases, strategic commitments, or whether a recent change already addressed the signal. Read the row as a prompt to inspect its source.

Know what each recommendation value counts

These values do not share one unit. Ten crawl findings, ten lost positions, and ten opportunities are not equivalent impacts. The thin bar only normalizes a row against the largest number among the visible recommendations. It is not an ROI, severity, probability, or effort scale.

RecommendationValue and destination
Critical, high, medium, or low site findingsSeverity count for the primary site's usable completed crawl; row opens Site Audit
Strongest keyword lossAbsolute position delta of the leading loser in the selected list tracking result; row opens Position Tracking
High-priority opportunitiesCritical plus high result count from the workspace-wide opportunity evaluation; row opens Action Center
Missing own rankWorkspace evaluation count of keywords without own-rank evidence, shown only when a primary list context exists; row opens the Keyword Database

A recommendation does not prove the wording around it

The localized row sentence may use urgent language, but the underlying model supplies a count or rank delta, source family, tone, and destination. It does not diagnose crawling, visitor behavior, ranking loss, content gaps, or business harm. Those explanations remain hypotheses until the detailed run, URLs, SERPs, pages, and relevant change history support them.

For Site Audit, review the issue definition and affected URLs. For a keyword loss, compare schedule context, observed URL, SERP snapshots, and dates. For an opportunity, inspect its input metrics, missing evidence, and action rationale. Correct a misleading hypothesis before it becomes a ticket title.

Treat an empty queue as no visible recommendation, not all clear

Nothing urgent means the current builder produced no rows inside its available sources and policy limit. That can happen because relevant severity counts are zero, no comparable loss exists, the opportunity layer returned no critical or high item, or missing-rank evidence is zero. It can also happen because data, permission, list context, or a suitable completed run is absent.

The panel does not test every SEO risk, commercial goal, unresolved operation, low-confidence metric, or page-level concern. Verify setup and the owning surfaces before reporting that the workspace has no work.

Home does not currently show a recent-activity feed

The current Home footer renders Activity as a compact tool link. When keyword activity events exist, the link can show the start date of the newest such event. When none exist, it shows an invitation to open Activity. It does not display event kind, status, counters, failure evidence, cost, support code, or an event list.

The consolidated backend summary can assemble bounded activity rows from Site Audit runs, domain snapshots, keyword discovery, and keyword activity. Home does not render that array. Therefore the footer date is not the latest cross-tool operation and must not be described as a complete recent workspace history.

Use the Activity page for operation evidence

Open Activity when you need statuses, counters, partial or failed outcomes, pagination, notifications, or older recorded history. Then open the feature that created the operation for its current job, result, retry, cancellation, schedule, or cost state. An Activity event and a live job record answer different questions.

Absence from Home proves nothing about whether a crawl, domain snapshot, research run, import, enrichment, or tracking operation happened. Even absence from Activity requires checking workspace, loaded history, event coverage, and the owning feature before concluding that no operation exists.

Keep the newest schedule summary separate from selected-list movers

Your rankings uses the newest active tracking schedule, its measured average-position trend, current position bands, schedule label, cadence, keyword count, last run, and available run coverage. Additional active schedules are counted but not charted. No active schedule produces setup; too little measured history produces no trend.

What moved uses the selected list tracking dashboard and shows the leading gains and losses for that list. The selected list defaults to the largest one in the bounded list set and can differ from the newest active schedule. Never combine the chart and movers into one narrative until list or selected keywords, domain, market, language, device, depth, and measurement dates match.

Follow the row destination, not only the panel header

The panel header links to Action Center, but individual rows route according to their evidence. Site findings open Site Audit. A loss opens Position Tracking. An opportunity opens Action Center. Missing own rank opens the Keyword Database. The header is a general shortcut, not proof that Action Center owns every visible recommendation.

1
Open the row destination and reproduce its exact scope, source, and date.
2
Check whether the visible value is complete, capped, unknown, or based on a selected list or primary site.
3
Write the hypothesis separately from the observed fact.
4
Confirm owner, proposed change, constraints, and expected effect with a person.
5
Define the next comparable crawl, tracking run, snapshot, or evaluation before implementation.
6
Verify the operation state in Activity where supported and the outcome in the owning evidence.

Do not confuse completion with impact

A completed crawl can still expose unresolved issues. A completed tracking run can have partial coverage. A completed opportunity evaluation can validly return zero. A completed implementation task can fail to improve the intended metric. Home does not join these stages into one causal chain.

Keep request, operation, output, human decision, implementation, and comparable follow-up measurement as separate records. Only the last comparison can support an outcome claim, and even then it may show correlation rather than cause.

Run a short review without turning Home into a task system

1
Confirm workspace, permissions, selected list, newest active schedule, primary Site Audit site, and loading state.
2
Read the three visible recommendation rows and note which valid lower rows may be outside the cap.
3
Open the owner for any new or materially changed recommendation.
4
Open Activity separately for failures, partial outcomes, or operations that need reconciliation.
5
Assign work only after rationale, owner, constraint, and verification measure are explicit.
6
Leave static summaries alone when no new evidence can change the decision.