Skip to content

Chapter 10: Reading Your Reports

Audience: media buyers, campaign managers

This chapter explains the analytics panels across Cat-Scan and how to interpret the numbers.

Spend stats

Available on the home page and in config drill-downs.

Metric What it tells you
Total spend Gross spend across the selected period and seat.
Spend trend Recent period vs. previous period. Rising spend with flat wins = cost inflation.
Spend by config Which pretargeting config is responsible for how much spend. Helps identify which configs to optimize first.

Config performance

Shows how each pretargeting config performed over time.

  • Daily breakdown: per-config impressions, clicks, spend, win rate, CTR, and CPM over the selected period.
  • Trend lines: spot configs whose performance is degrading.
  • Field breakdown: which specific fields (geos, sizes, formats) within a config are driving the numbers.

Endpoint efficiency

Shows QPS utilization per bidder endpoint.

  • Efficiency ratio: useful QPS / total QPS. Closer to 1.0 is better.
  • Per-endpoint breakdown: if your bidder has multiple endpoints, see which ones are most and least efficient.
  • Use this to decide whether consolidating endpoints would help.

Snapshot comparisons

After rolling back a pretargeting change (or applying a new one), the snapshot comparison panel shows:

  • Before: config state prior to the change
  • After: config state after the change
  • Delta: what exactly changed (fields added/removed/modified)

This is useful for post-change analysis: "I excluded 5 geos yesterday, so what happened to my funnel?"

Cat-Scan may display AI-generated recommendations based on your data. These suggest specific config changes with estimated impact. They are suggestions, not automatic actions. You always choose whether to apply them.

Tips for reading reports

  1. Always check the period selector. A 7-day view and a 30-day view can tell very different stories.
  2. Compare configs, don't just look at totals. One bad config can drag down aggregate numbers while other configs perform well.
  3. Look at trends, not snapshots. A single day's data is noisy. Trends over 7-14 days are more reliable.
  4. Cross-reference dimensions. High waste in geo view + high waste in size view for the same config = two separate optimization opportunities.