Google Ads Measurement cheat sheet

Conversion tracking prerequisite: the Google Ads conversion action and the GA4 conversion event must be linked through the same Google Ads account; unlinked conversions appear in Ads only as imported or as Ads-native actions.

Auto-tagging (gclid) must be enabled in Google Ads before GA4 can attribute paid Google traffic; manual UTM tagging alone leaves gclid absent and breaks Ads-to-Analytics session stitching.

GA4 Google Ads link requires Admin > Product links > Google Ads links, with the linking user holding edit permission on both the GA4 property and the Ads account.

Attribution model is a property-level setting in GA4 (Admin > Attribution settings); changing it reprocesses historical data, while Google Ads conversion actions each carry their own attribution model.

Conversion window defaults: GA4 acquisition reports use 30 days; Google Ads conversion actions default to 30-day click and 1-day view unless edited per action.

Data-driven attribution is the default for GA4 reports and for new Google Ads conversion actions; last-click remains selectable but is no longer the default.

Remarketing audiences built in GA4 require the Google Signals or User-ID setup plus the Google Ads link; audience membership populates only after the link is active.

Consent Mode v2 affects modeled conversions: without consent signals, GA4 drops ad_storage/analytics_storage identifiers and Google Ads reports modeled or unobserved conversions instead of measured ones.

Enhanced conversions match hashed first-party identifiers (email, phone, address) to Google accounts; they require the conversion tag to capture user-provided data and must be enabled per conversion action.

CTR = clicks / impressions; CVR = conversions / clicks; CPM = (cost / impressions) x 1000; CPC = cost / clicks; CPA = cost / conversions; ROAS = conversion value / cost.

Conversion rate in GA4 = conversions / sessions for the selected event set, so adding events to the key-event list mechanically changes reported rates.

Sessions in GA4 are not closed by a date boundary; a session crossing midnight is attributed whole to its start date, which can shift day-level comparisons against Google Ads.

Engagement rate = engaged sessions / sessions, where an engaged session lasts 10+ seconds, fires a conversion event, or views 2+ pages.

Google Ads 'Conversions' column counts only actions marked Primary; 'All conversions' adds Secondary actions, so the two columns are not interchangeable.

Impression share = impressions / eligible impressions; lost IS (budget) plus lost IS (rank) plus impression share equals 100% of eligible impressions.

GA4 data freshness: standard reports refresh within 24-48 hours, real-time within seconds; Google Ads conversion reporting can lag up to 3 hours for the Conversions column.

Thresholding in GA4 hides rows or withholds demographic detail when user counts are too low to protect identity; missing rows are a privacy artifact, not zero traffic.

Sampling: GA4 standard reports are unsampled for most cards, but Explorations and the API apply sampling or cardinality limits that change totals versus the UI.

Budget-limited campaigns show lost IS (budget) above zero and impression share below 100%; raising budget is the lever, not bid changes.

Rank-limited campaigns show lost IS (rank); levers are bid, Quality Score components (expected CTR, ad relevance, landing page experience), and ad/asset relevance.

Smart Bidding target changes need 1-2 conversion cycles to stabilize; evaluate after the learning period rather than within days of an edit.

tCPA and tROAS bid to the target on average, so segments (device, geo, audience) can deviate widely while the account-level target holds.

Search terms report drives negatives and match-type changes; broad match plus smart bidding requires negative keyword lists and account-level brand exclusions to control spend.

Value-based bidding requires conversion values passed via the tag or imported from Analytics; without values, tROAS and Maximize conversion value cannot optimize.

Ad strength and asset ratings (Low/Good/Best) indicate coverage of headlines and descriptions; Responsive Search Ads need at least 3 headlines and 2 descriptions to serve.

Experiment structure: change one lever per test (bid strategy, budget, landing page), hold the audience and creative constant, and use Google Ads drafts/experiments for a clean split.

GA4 Explorations (funnel, path, segment overlap) convert insight into action by exposing drop-off steps; the resulting segment can be exported to Google Ads as an audience.

Conversion tracking prerequisite: the Google Ads conversion action and the GA4 conversion event must be linked through the same Google Ads account; unlinked conversions appear in Ads only as imported or as Ads-native actions.

Auto-tagging (gclid) must be enabled in Google Ads before GA4 can attribute paid Google traffic; manual UTM tagging alone leaves gclid absent and breaks Ads-to-Analytics session stitching.

GA4 Google Ads link requires Admin > Product links > Google Ads links, with the linking user holding edit permission on both the GA4 property and the Ads account.

Attribution model is a property-level setting in GA4 (Admin > Attribution settings); changing it reprocesses historical data, while Google Ads conversion actions each carry their own attribution model.

Conversion window defaults: GA4 acquisition reports use 30 days; Google Ads conversion actions default to 30-day click and 1-day view unless edited per action.

Data-driven attribution is the default for GA4 reports and for new Google Ads conversion actions; last-click remains selectable but is no longer the default.

Remarketing audiences built in GA4 require the Google Signals or User-ID setup plus the Google Ads link; audience membership populates only after the link is active.

Consent Mode v2 affects modeled conversions: without consent signals, GA4 drops ad_storage/analytics_storage identifiers and Google Ads reports modeled or unobserved conversions instead of measured ones.

Enhanced conversions match hashed first-party identifiers (email, phone, address) to Google accounts; they require the conversion tag to capture user-provided data and must be enabled per conversion action.

CTR = clicks / impressions; CVR = conversions / clicks; CPM = (cost / impressions) x 1000; CPC = cost / clicks; CPA = cost / conversions; ROAS = conversion value / cost.

Conversion rate in GA4 = conversions / sessions for the selected event set, so adding events to the key-event list mechanically changes reported rates.

Sessions in GA4 are not closed by a date boundary; a session crossing midnight is attributed whole to its start date, which can shift day-level comparisons against Google Ads.

Engagement rate = engaged sessions / sessions, where an engaged session lasts 10+ seconds, fires a conversion event, or views 2+ pages.

Google Ads 'Conversions' column counts only actions marked Primary; 'All conversions' adds Secondary actions, so the two columns are not interchangeable.

Impression share = impressions / eligible impressions; lost IS (budget) plus lost IS (rank) plus impression share equals 100% of eligible impressions.

GA4 data freshness: standard reports refresh within 24-48 hours, real-time within seconds; Google Ads conversion reporting can lag up to 3 hours for the Conversions column.

Thresholding in GA4 hides rows or withholds demographic detail when user counts are too low to protect identity; missing rows are a privacy artifact, not zero traffic.

Sampling: GA4 standard reports are unsampled for most cards, but Explorations and the API apply sampling or cardinality limits that change totals versus the UI.

Budget-limited campaigns show lost IS (budget) above zero and impression share below 100%; raising budget is the lever, not bid changes.

Rank-limited campaigns show lost IS (rank); levers are bid, Quality Score components (expected CTR, ad relevance, landing page experience), and ad/asset relevance.

Smart Bidding target changes need 1-2 conversion cycles to stabilize; evaluate after the learning period rather than within days of an edit.

tCPA and tROAS bid to the target on average, so segments (device, geo, audience) can deviate widely while the account-level target holds.

Search terms report drives negatives and match-type changes; broad match plus smart bidding requires negative keyword lists and account-level brand exclusions to control spend.

Value-based bidding requires conversion values passed via the tag or imported from Analytics; without values, tROAS and Maximize conversion value cannot optimize.

Ad strength and asset ratings (Low/Good/Best) indicate coverage of headlines and descriptions; Responsive Search Ads need at least 3 headlines and 2 descriptions to serve.

Experiment structure: change one lever per test (bid strategy, budget, landing page), hold the audience and creative constant, and use Google Ads drafts/experiments for a clean split.

GA4 Explorations (funnel, path, segment overlap) convert insight into action by exposing drop-off steps; the resulting segment can be exported to Google Ads as an audience.