Google Ads Measurement
This domain tests how conversions are counted, attributed, and reported. Expect questions on conversion actions, counting settings, attribution models, conversion windows, enhanced conversions, and consent - and on why two reports can disagree.
Core rules for measurement
- Conversion actions can be primary or secondary. “Count one” records one conversion per click; “count every” records each conversion after a click. A purchase usually uses count one; a newsletter sign-up may use count every.
- Click-through conversion windows default to 30 days and can be set up to 60 days. View-through windows default to 1 day and can be set up to 30 days. Video engaged-view conversions are reported separately.
- Attribution models include last click, first click, linear, time decay, position-based, and data-driven. Data-driven attribution distributes credit across touchpoints using modeled paths; last click gives all credit to the final click.
- Enhanced conversions use hashed first-party data, such as email or phone, to improve match rates. Consent Mode adjusts tag behavior for consent and enables modeled conversions.
- Imported GA4 conversions and native Google Ads conversions can double count if both are used carelessly.
Worked example
An ecommerce store records 100 purchases. Last-click attribution gives Search 40, Display 5, Email 15, and Direct 40. Data-driven attribution gives Search 55, Display 15, Email 20, and Direct 10. The shift occurs because Display and Email often assist earlier in the path, while last click credits the final touch. If Search uses a 30-day click window but Display uses a 1-day view-through window, Display may still be undercredited. The team aligns windows with the sales cycle, uses data-driven attribution for budget decisions, and keeps last click for operational comparison.
Traps that cost marks
- Using “count every” for a one-time purchase, inflating conversions.
- Comparing models as if they were interchangeable rather than different credit rules.
- Forgetting that view-through and engaged-view conversions follow different windows.
- Double counting imported and native conversions.
- Ignoring consent and enhanced conversions when match rates are low.