Media

Performance Anomaly Watcher

Cost-per-acquisition spikes or click-through-rate drops. Slack alert with diagnosis and a suggested fix within 30 minutes.

Meta AdsGoogle AdsSlackLooker Studio
Performance Anomaly Watcher

Typical workflow

A Meta Ads campaign goes live on Monday. The watcher pulls data every 30 minutes via the Meta Marketing API. On Wednesday afternoon, the CPA on one ad set climbs 35% above the 14-day average. Within 30 minutes, the media buyer gets a Slack message identifying the ad set, showing the CPA trend, and suggesting a bid cap adjustment or creative swap. The fix takes 5 minutes instead of a morning audit.

How it works

1

Step 1

Checks live campaign data every 30 minutes against your 14-day historical baseline for cost per acquisition (CPA), click-through rate (CTR), and return on ad spend (ROAS).

2

Step 2

Triggers an alert when any metric moves more than 20% outside the baseline, along with the ad set and creative responsible.

3

Step 3

Posts a Slack message with the diagnosis, the affected campaign, and a 3-bullet response memo the media buyer can act on immediately.

Why this is not ChatGPT

  • ✓
    It runs on a 30-minute loop against live API data. It catches a CPA spike in under an hour, not the next morning.
  • ✓
    It compares against your own campaign baseline, not industry averages. A $10 CPA may be fine for you and terrible for another business.
  • ✓
    It writes a specific response memo tied to the exact ad set and creative that triggered the problem, not a generic "check your targeting" suggestion.

Typical Build

12days

Best fit for

Growth teams and media agencies running always-on paid campaigns with daily spend above 100 USD. Works best when you have at least 14 days of campaign history to build a reliable baseline.