Assess for Success: Marketing Analytics and Measurement
Domain 5, Assess for Success: Marketing Analytics and Measurement, tests whether you can choose metrics, track campaigns, interpret results, and turn data into decisions. Expect questions on KPIs, dimensions and metrics, conversion tracking, attribution, ROI, ROAS, and testing.
Core rules and concepts
- Start with a measurement plan: business objective, question, KPI, metric, data source, and owner. A KPI is a few decision-critical measures; a metric is any quantifiable value.
- In Google Analytics 4, events describe actions, and conversions are important events. Dimensions describe data, such as channel or device; metrics are numbers, such as sessions or revenue. Use UTM parameters to identify campaign source, medium, and name.
- Conversion tracking must be accurate and consistent. Missing tags, duplicate purchases, or bot traffic can make a campaign look better or worse than it is.
- Attribution assigns credit to touchpoints. Last-click gives all credit to the final interaction; it is simple but undervalues earlier awareness. Compare models and use holdouts or experiments when possible.
- ROAS is revenue divided by ad spend. ROI is profit divided by cost, usually expressed as a percentage. They are not interchangeable. Statistical significance helps avoid acting on random noise.
Worked example
A campaign spends 2,000 and generates 8,000 in revenue from 1,000 purchases across 50,000 sessions. ROAS is 8,000 divided by 2,000, or 4. Conversion rate is 1,000 divided by 50,000, or 2 percent. Cost per acquisition is 2,000 divided by 1,000, or 2. If the campaign's total cost including product and fulfillment is 6,000, profit is 2,000 and ROI is 2,000 divided by 6,000, or about 33 percent. The team checks whether the lift came from the campaign or from seasonality before increasing budget.
Traps that cost marks
Confusing ROAS with ROI; reporting sessions when the question asks about users; trusting last-click as the whole story; ignoring tracking gaps; and calling a small difference a winner without enough data.