Sequential data preparation → parallel specialist evidence collection → sequential final decision synthesis.
Sequential setup → parallel specialist fan-out → final decision fan-in.
All agents completed. Refresh the dashboard to load the latest result from this run.
No summary available.
Compares the forecast created from the previous checkpoint with actual metrics from the latest checkpoint.
meta.window_end equal to or newer than forecast_for_date.
Composite score from activation, retention, monetization, and release risk.
Evidence layers consumed by the Final Decision Agent. This is not a debate transcript or a step-by-step sequence; specialists are peer inputs and the final agent performs fan-in synthesis.
| Layer | Signal | Decision Weight | Detail |
|---|---|---|---|
|
Item 1
Activation Agent
|
Activation Agent: The add-food flow is not the main bottleneck. Once users reach the workspace, food_add_success looks reasonably healthy; the problem is getting sessions into the workspace.
INVESTIGATE
|
The decision focus shifts to improving workspace entry and onboarding/home CTA. | |
|
Evidence
- Constraint / Counter-signal
Rejects blaming the add-food feature as the main root cause. Full Decision Weight
The decision focus shifts to improving workspace entry and onboarding/home CTA. |
|||
|
Item 2
Monetization Agent
|
Monetization Agent: There is an early revenue signal, but the purchase sample is small and a global paywall risks appearing too early.
SEGMENT ONLY
|
Monetization is not turned off, but limited to cohorts that already reached the value moment. | |
|
Evidence
- Constraint / Counter-signal
Keeps the revenue opportunity testable, but only through segmentation. Full Decision Weight
Monetization is not turned off, but limited to cohorts that already reached the value moment. |
|||
|
Item 3
Retention Agent
|
Retention Agent: I caution against aggressive scaling because D1/habit is still weak; extra traffic will leak before becoming a habit.
CAUTION AGAINST SCALE
|
Ads are held, and product priority shifts to retention and the D0-D1 loop. | |
|
Evidence
- Constraint / Counter-signal
Constrains ads/paywall scaling before D1 logged rate improves. Full Decision Weight
Ads are held, and product priority shifts to retention and the D0-D1 loop. |
|||
|
Item 4
Version Agent
|
Version Agent: Release does not look like the main risk source; rollout can continue gradually with guardrails.
CONTINUE WITH MONITORING
|
The final decision does not choose rollback; focus stays on retention and monetization timing. | |
|
Evidence
- Constraint / Counter-signal
- Full Decision Weight
The final decision does not choose rollback; focus stays on retention and monetization timing. |
|||
|
Item 5
Ads Agent
|
Ads Agent: I read acquisition performance and campaign lifecycle. The old Volume Stabil campaign should not automatically be treated as the main campaign if it is marked degraded legacy and Volume Install Reset exists as the reset successor.
MONITOR ADS
|
Ads become supporting evidence for the budget decision. | |
|
Evidence
- Constraint / Counter-signal
Do not interpret pausing/reducing Volume Stabil as shutting down acquisition if the reset campaign is being evaluated. Full Decision Weight
Ads become supporting evidence for the budget decision. |
|||
|
Item 6
Tomorrow Forecast Agent
|
Tomorrow Forecast Agent: Tomorrow forecast strengthens today decision guardrails; scaling is allowed only if predicted activation/retention risk stays safe.
FORECAST ALLOWS CAUTIOUS TEST
|
Forecast becomes a forward-looking guardrail for today decision. | |
|
Evidence
Forecast date: -; scaling guardrail: -; main risk: - Constraint / Counter-signal
Forecast does not block small experiments, but still requires actual evaluation tomorrow. Full Decision Weight
Forecast becomes a forward-looking guardrail for today decision. |
|||
|
Item 7
Final Decision Agent
|
Final Decision Agent: The retention constraint is stronger than monetization upside; today decision is Hold & Optimize, not aggressive scaling.
HOLD AND OPTIMIZE
|
- | |
|
Evidence
- Constraint / Counter-signal
Combines revenue signal, release safety, and retention guardrail into one operating decision. Full Decision Weight
- |
|||
Action plan that can be executed and evaluated objectively.
Operational action plan is not available yet. Clear cache and regenerate the analysis after the FinalDecisionAgent patch.
Evaluation loop: did the previous checkpoint decision prove correct based on the current outcome?
Summary of 6 specialist agents. Click a card to open its analysis detail.
Quantitative forecast based on the latest available data. Evaluation runs only when actual data for the forecast date is available.
-
-
-
-
Reads Google Ads performance, campaign lifecycle, and reset campaign context so the budget decision is not misinterpreted.
{
"meta": {
"app_name": "AI Growth Doctor",
"window_start": null,
"window_end": null,
"analyzed_at": null,
"architecture": "async_multi_agent_progress",
"run_id": null
},
"workflow": [],
"interaction_log": [],
"metrics": {
"activation_metrics": [],
"retention_metrics": [],
"monetization_metrics": [],
"version_metrics": [],
"ads_metrics": [],
"tomorrow_forecast_metrics": [],
"rule_based_decision": [],
"guardrail_policy": []
},
"app_profile": {
"tenant_id": "tenant_demo_001",
"app_id": "hitung_kalori_case_study",
"app_name": "Hitung Kalori",
"app_category": "health_fitness",
"core_action_name": "food logging",
"core_action_success_label": "food_add_success",
"workspace_name": "diary workspace",
"monetization_model": "subscription",
"timezone": "Asia/Jakarta",
"data_mode": "real_aggregated_case_study"
},
"metric_mapping": {
"activation.entry_users": "activation_metrics.metrics_7d.session_users",
"activation.workspace_users": "activation_metrics.metrics_7d.workspace_users",
"activation.core_action_success_users": "activation_metrics.metrics_7d.food_add_success_users",
"activation.core_action_success_rate_from_entry": "activation_metrics.metrics_7d.food_add_success_rate_from_session",
"activation.core_action_success_rate_from_workspace": "activation_metrics.metrics_7d.food_add_success_rate_from_workspace",
"retention.d0_rate": "retention_metrics.metrics_7d_avg.d0_logged_rate",
"retention.d1_rate": "retention_metrics.metrics_7d_avg.d1_logged_rate",
"retention.habit_7d_rate": "retention_metrics.metrics_7d_avg.habit_7d_rate",
"retention.avg_active_days_7d": "retention_metrics.metrics_7d_avg.avg_log_days_7d",
"monetization.exposure_users": "monetization_metrics.metrics_7d.paywall_view_users",
"monetization.purchase_start_users": "monetization_metrics.metrics_7d.purchase_start_users",
"monetization.purchase_success_users": "monetization_metrics.metrics_7d.purchase_success_users",
"monetization.purchase_success_rate_from_exposure": "monetization_metrics.metrics_7d.purchase_success_rate_from_paywall",
"ads.cost": "ads_metrics.overall.cost",
"ads.clicks": "ads_metrics.overall.clicks",
"ads.impressions": "ads_metrics.overall.impressions",
"ads.conversions": "ads_metrics.overall.conversions",
"ads.cost_per_conversion": "ads_metrics.overall.cost_per_install",
"ads.conversion_rate": "ads_metrics.overall.conversion_rate"
},
"generic_metrics_context": [],
"mapping_validation": [],
"source_metric_refs": [],
"evaluations": {
"forecast_evaluations": {
"status": "empty",
"evaluated": [],
"pending": [],
"skipped": []
},
"forecast_model_calibration": {
"status": "empty",
"evaluations_used": 0,
"trust_score": {
"updated_score": null,
"interpretation": "empty"
},
"decision_instruction": []
}
},
"agents": {
"activation_agent": [],
"retention_agent": [],
"monetization_agent": [],
"decision_agent": [],
"ai_decision_agent": [],
"ai_activation_agent": [],
"ai_retention_agent": [],
"ai_monetization_agent": [],
"ai_version_agent": [],
"ai_ads_agent": [],
"ai_tomorrow_forecast_agent": [],
"structured_negotiation": [],
"orchestrator_evidence_assembly": [],
"ai_final_decision_agent": [],
"decision_scenario_simulator": []
},
"structured_negotiation": [],
"conflict_matrix": [],
"negotiation_summary": [],
"orchestrator_evidence_assembly": [],
"quantitative_baseline_comparison": [],
"charts": {
"activation_trend": [],
"retention_trend": []
}
}