Skip to main content

Growth Metrics Overview

We assess the reliability of panel-based datasets using four core confidence dimensions. These measures help determine how closely our panel data reflects true market behavior, both in scale and trend consistency.

1. Absolute Volume​

Measures how much of an entity’s actual spend or engagement is captured in the panel, compared to a ground truth source.

Example: How much of a brand’s reported revenue is reflected in total receipt spend in our transaction panel?

This helps answer:

  • Does this volume scale vary across brands?
  • How does it shift over time or across regions?

2. Relative Trend – Correlation​

Assesses the correlation between an entity’s changes in spend or engagement over time and space, compared to a trusted reference source.

A high correlation score indicates strong alignment with ground truth data, suggesting high confidence — even if absolute volume is lower.

3. Relative Trend – Predictability​

Measures how accurately panel-based metrics can predict reference values over time.

  • Uses historical data to perform a regression of reference volume change on panel volume change.
  • Confidence is evaluated using Mean Absolute Percent Error (MAPE).

4. Relative Trend – Directional Correctness​

Evaluates whether the second derivative (i.e., change in growth rate) of the panel data moves in the same direction as the reference dataset.

Panel Variance (Stability Sub-Metric)​

In addition to directional accuracy, we assess the stability of volume coverage by measuring variance:

  • Low variance: Consistent coverage → higher confidence
  • High variance: Volatile coverage → lower confidence and poor model fit

Together, these metrics guide our internal panel QA and give users visibility into how much trust to place in trend movements across brands, geographies, and time.