Correlation-Based Performance Insights in Copilot
Updated
Overview
Marketers often notice sudden performance spikes or drops but struggle to understand why they happened. Traditional dashboards highlight what changed, but they rarely explain the underlying drivers. As a result, teams must manually investigate budgets, creatives, audiences, and delivery changes—leading to guesswork, slow decisions, and delayed optimization.
Correlation‑Based Performance Insights removes this friction. Copilot correlates reporting metrics with audit‑level changes across budgets, bids, audiences, creatives, and runtime behavior to explain why performance shifted. It delivers a structured root‑cause analysis with quantified impact scores, helping teams quickly identify key drivers and take confident, data‑backed action.
How Correlation-Based Performance Insights Helps
Copilot's correlation-based performance insights helps you quickly uncover the “why” behind performance changes using advanced diagnostic intelligence.
Key Benefits
Pinpoint likely causes quickly: Copilot analyzes patterns across budgets, audiences, creatives, and delivery to explain performance drops or spikes.
Context‑aware insights: Explanations adapt to your selected campaigns, metrics, and date range, ensuring higher relevance.
Transparent insights trace: You can clearly see which factors influenced each explanation, making the reasoning easy to follow and trust.
Audit‑change signals considered: Copilot detects key changes such as budget updates, bid adjustments, creative swaps, and audience modifications.
Performance data validation: The system cross‑checks explanations against reporting data—for example, identifying whether reduced spend occurred in high‑performing regions.
Multi‑entity analysis: Copilot evaluates impact across campaigns, ad sets, creatives, geographies, and channels, avoiding narrow or misleading conclusions.
Source‑backed explanations: Each insight clearly shows the data points and entities used, ensuring explanations remain evidence‑driven and transparent.
Factors Included in Correlation-Based Performance Insights
When you ask a causal question in Copilot, the system analyzes your prompt and evaluates metric movement, contributing entities, and audit‑level changes to generate deep performance insights. Copilot generates these insights by examining three core categories of factors that together surface the most likely drivers.
System Factors
System factors represent advertiser‑controlled changes that directly influence delivery and performance. Copilot detects and evaluates changes such as:
Budget updates
Bid or bidding strategy adjustments
Targeting changes
Creative updates
Campaign runtime changes
Using these changes, Copilot measures the magnitude of the performance drop or spike—both in absolute and percentage terms—and identifies the affected metric, such as impressions, CTR, or conversions.
Hierarchical Factors
Hierarchical factors help Copilot identify where the performance change originated within the campaign structure. The system pinpoints the specific initiatives and entities driving the shift by breaking down metrics across:
Campaign name
Ad set name
Ad variant name
Channel
This analysis ensures Copilot attributes impact to the correct level of the advertising hierarchy instead of relying on high‑level averages.
Segmentation Factors
Segmentation factors explain the performance change across audiences and delivery contexts. Copilot breaks down metrics using key dimensions such as:
Age
Gender
Country
Device
Placement
Date range
By combining these segment‑level insights with system and hierarchical factors, Copilot delivers a complete, context‑aware explanation of performance changes.
Correlation‑Based Performance Insights helps teams move from observation to action. By identifying what actually drove a performance change and ranking the most impactful factors, Copilot shows exactly where to focus optimization efforts. Because the analysis accounts for timing, context, and cross‑entity effects, the insights are consistent, reliable, and free from misleading assumptions. This goes beyond traditional dashboards or rule‑based alerts, delivering a fundamentally new analytical capability that teams can trust to make faster, smarter decisions.