skill-010-campaign-priority-experiments

name	skill-010-campaign-priority-experiments
description	Ensure campaign priorities (especially for Shopping) are correctly configured and that experiments/drafts are properly structured.
allowed-tools	Read, Grep, Glob

Skill 010: Campaign Priority and Experiment Structure

Purpose

Ensure campaign priorities (especially for Shopping) are correctly configured and that experiments/drafts are properly structured. Incorrect priority settings cause traffic to route to the wrong campaigns, distorting results.

Data Requirements

Data Source: Custom GAQL Required

Standard export includes basic shopping settings but may lack complete priority data. Custom GAQL provides experiment information.

GAQL Query (Shopping priorities):

SELECT
  campaign.id,
  campaign.name,
  campaign.status,
  campaign.shopping_setting.campaign_priority,
  campaign.shopping_setting.merchant_id,
  campaign.advertising_channel_type
FROM campaign
WHERE campaign.advertising_channel_type = 'SHOPPING'
  AND campaign.status != 'REMOVED'

GAQL Query (experiments):

SELECT
  experiment.experiment_id,
  experiment.name,
  experiment.status,
  experiment.type,
  experiment.start_date,
  experiment.end_date
FROM experiment

Run via /google-ads:get-custom with query names shopping_priorities and experiments.

Analysis Steps

1. Analyze Shopping priorities: List all Shopping campaigns with priorities (0=Low, 1=Medium, 2=High); check for conflicts
2. Validate priority logic: Higher priority campaigns should have negatives to funnel specific queries
3. Review experiments: List active experiments; check duration (2-4 weeks recommended) and traffic split
4. Check experiment best practices: One variable tested, sufficient budget, clear success metrics
5. Identify issues: Overlapping priorities without differentiation, experiments running too long/short

Thresholds

Condition	Severity
Multiple Shopping campaigns with same priority and overlapping products	Critical
High priority campaign without negatives	Warning
Experiment running > 8 weeks	Warning
Experiment < 2 weeks	Info
Experiment traffic split uneven (>60/40)	Info

Output

Use Short format by default. Use Detailed if user requests comprehensive analysis.

Short:

## Campaign Priority & Experiments Audit
**Account:** [Name] | **Shopping Campaigns:** [X] | **Active Experiments:** [Y]

### Critical ([Count])
- **[Campaign A] & [Campaign B]**: Same priority (Medium), overlapping products → Differentiate priorities

### Warnings ([Count])
- **[Campaign]**: High priority without negatives → Add negatives to filter queries
- **[Experiment]**: Running [X] weeks (>8 weeks) → Conclude and implement winner

### Recommendations
1. Resolve priority conflicts in Shopping campaigns
2. Conclude long-running experiments

Detailed adds:

• What Was Checked (priority structure, negative keywords, experiment duration)
• Shopping priority table (Campaign, Priority, Products, Negatives)
• Experiment status table (Experiment, Base, Duration, Split, Status)
• Recommended priority structure