Ask a more precise question of your data. These prompts cover retention, experimental validity, metric trees, survey bias and dashboard design. Describe how the data was collected and separate a statistical association from a causal conclusion.
How to use these prompts
Choose a template and open Preview & customise.
Fill the named fields with your details and check the completed draft.
Copy it into your preferred AI tool, or sign in to prepare a Chat draft in Ortivo.
Copying is free and does not call an AI. Generating an answer requires your own configured provider or an available trial. Review important facts and every draft before use.
Analyze the cohort retention pattern in {{cohort_data_summary}} for users acquired via {{acquisition_channel}} on a {{billing_cycle}} cycle. Identify the inflection point where re…
Analyze the cohort retention pattern in {{cohort_data_summary}} for users acquired via {{acquisition_channel}} on a {{billing_cycle}} cycle. Identify the inflection point where retention flattens, estimate steady-state LTV, and highlight early-month cliff drops.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.
Data & analytics
A/B Test Statistical Power & Validity Audit
Evaluate an experiment with baseline conversion {{baseline_rate}}, target lift {{minimum_detectable_effect}}, sample size {{sample_size}}, and duration {{test_duration}}. Check fo…
Suggested: Multi-Model CompareRead template
Evaluate an experiment with baseline conversion {{baseline_rate}}, target lift {{minimum_detectable_effect}}, sample size {{sample_size}}, and duration {{test_duration}}. Check for underpowered sizing, peeking/p-hacking risks, novelty effects, and Day-of-Week bias.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.
Data & analytics
North Star Metric & KPI Driver Tree Builder
Decompose {{north_star_metric}} for our {{business_model}} into a 3-level mathematical driver tree. Assign leading input metrics to {{team_departments}}, and identify potential Go…
Suggested: Multi-Model CompareRead template
Decompose {{north_star_metric}} for our {{business_model}} into a 3-level mathematical driver tree. Assign leading input metrics to {{team_departments}}, and identify potential Goodhart's Law perverse incentives if any single sub-metric is over-optimized.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.
Data & analytics
Survey Instrument Bias & Question Polisher
Audit these draft survey questions ({{draft_questions}}) aimed at {{target_respondents}} to measure {{research_goal}}. Flag leading phrasing, double-barreled questions, and scale …
Suggested: Single AIRead template
Audit these draft survey questions ({{draft_questions}}) aimed at {{target_respondents}} to measure {{research_goal}}. Flag leading phrasing, double-barreled questions, and scale imbalances, and rewrite them for clean quantitative analysis.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.
Data & analytics
Causal Inference vs. Spurious Correlation Check
We observed {{observed_correlation}} and plan to execute {{proposed_intervention}}. Evaluate how {{confounding_factors}}, reverse causality, or selection bias could explain this p…
Suggested: AI CouncilRead template
We observed {{observed_correlation}} and plan to execute {{proposed_intervention}}. Evaluate how {{confounding_factors}}, reverse causality, or selection bias could explain this pattern, and design a quasi-experiment (diff-in-diff or holdout) to prove causation.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.
Data & analytics
Time-Series Anomaly & Seasonality Attribution
Investigate the spike/drop in {{metric_name}} during {{anomaly_window}} alongside {{concurrent_events}}. Separate cyclical seasonality and tracking instrumentation changes from ge…
Investigate the spike/drop in {{metric_name}} during {{anomaly_window}} alongside {{concurrent_events}}. Separate cyclical seasonality and tracking instrumentation changes from genuine behavioral shifts, and list verification queries to run.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.
Data & analytics
Predictive Churn Model Feature Engineering
Using {{available_telemetry}}, propose the top 10 behavioral velocity and ratio features to predict churn within {{prediction_window}} for {{customer_type}}. Flag any features at …
Suggested: Multi-Model CompareRead template
Using {{available_telemetry}}, propose the top 10 behavioral velocity and ratio features to predict churn within {{prediction_window}} for {{customer_type}}. Flag any features at risk of target leakage and explain how to operationalize the score for intervention.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.
Data & analytics
Data Warehouse Schema & Lineage Reviewer
Design a clean dimensional (star/snowflake) schema from {{raw_tables}} optimized to answer {{reporting_questions}} in {{bi_tool}}. Define fact tables, slowly changing dimensions (…
Suggested: Single AIRead template
Design a clean dimensional (star/snowflake) schema from {{raw_tables}} optimized to answer {{reporting_questions}} in {{bi_tool}}. Define fact tables, slowly changing dimensions (SCD Type 2), grain definitions, and data freshness assertions.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.
Data & analytics
Executive Dashboard Wireframe & Hierarchy Spec
Design a single-screen dashboard specification for {{audience_role}} reviewing performance on a {{decision_cadence}} cadence using {{available_metrics}}. Group metrics into Headli…
Suggested: Single AIRead template
Design a single-screen dashboard specification for {{audience_role}} reviewing performance on a {{decision_cadence}} cadence using {{available_metrics}}. Group metrics into Headline Health, Diagnostic Breakdowns, and Action Triggers, eliminating chart clutter.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.
Data & analytics
Price Elasticity & Van Westendorp Analyzer
Analyze {{pricing_survey_or_sales_data}} relative to {{competitor_anchors}} and unit marginal cost {{marginal_cost}}. Estimate the optimal price point, acceptable price range, and…
Suggested: Multi-Model CompareRead template
Analyze {{pricing_survey_or_sales_data}} relative to {{competitor_anchors}} and unit marginal cost {{marginal_cost}}. Estimate the optimal price point, acceptable price range, and where volume loss outweighs margin expansion.
Use only supplied material or sources actually accessed. Distinguish evidence, estimates and assumptions. Do not invent facts, citations, test results or actions performed. Flag missing information and tool limitations.