Are view-through conversions real or inflated?

They are partially real and largely inflated — a view-through conversion credits your ad for a conversion by a user who saw the ad (without clicking) then converted later, which captures some genuine awareness-driven influence but also claims credit for conversions that would have happened through organic search, direct traffic, or other channels regardless. Across audited accounts, 40–70% of view-through conversions fail a holdout test (the conversion happens without the ad being shown too). For direct-response campaigns, view-throughs are almost always double-counting organic demand; for awareness campaigns, a discounted share is reasonable. The Conversion Quality Audit shows your view-through percentage per campaign and recommends whether to weight them at 100%, a discounted share, or exclude them from bidding.

Should I include view-through conversions in my ROAS calculation?

Generally no for performance campaigns (Search, Shopping, PMax's shopping component) — view-throughs inflate reported ROAS in a way that doesn't reflect actual business value, and including them trains Smart Bidding toward traffic that does not incrementally convert. Include them, optionally at a discounted rate (say 30–50%), for awareness-focused campaigns (Display, YouTube) where the ad's job is to drive later intent. The rule of thumb: if you would report this ROAS to a CFO expecting causally-attributable revenue, exclude view-throughs. The Conversion Quality Audit recommends the view-through inclusion setting per campaign based on the campaign's role in your funnel.

How accurate are modeled conversions in Google Ads?

Google's internal benchmarks claim 70–90% accuracy on modeled conversions (conversions estimated rather than directly observed, used to fill gaps from iOS privacy, consent denials, and cross-device), but independent testing suggests real accuracy is closer to 50–70% in high-privacy markets. Modeled conversions scale up as consent mode denial rates rise — in GDPR-heavy markets they can account for 40–60% of Google Ads' reported conversion volume. This matters because Smart Bidding treats modeled conversions identically to verified ones when calibrating bids. The Conversion Quality Audit quantifies your modeled conversion share per campaign and flags campaigns where the share exceeds 30% as having degraded bidding signal.

How to tell if my conversions are being double-counted

Four diagnostic checks: (1) compare Google Ads conversion count against your backend order count for the same date range — if Google Ads exceeds backend by 20%+, something is double-firing, (2) check whether you have both GA4 conversion import AND native Google Ads tag active on the same event, (3) verify 'Count: One' vs 'Every' setting on each conversion action matches your business intent, (4) inspect network requests on your thank-you page for duplicate conversion pixel fires. Most double-counting traces to cause 2 or 3. The Conversion Quality Audit automates all four checks and flags the specific conversion actions with deduplication risk.

My conversion tracking is gaslighting me — how do I verify what's real?

Anchor on your backend order system (Shopify, WooCommerce, Stripe, CRM) as ground truth — that's the number you actually got paid for. Compare daily backend orders to Google Ads' attributed conversions and GA4's attributed conversions. If Google Ads says 500 and backend says 200, the 300 gap is some mix of view-through, modeled, cross-device, and double-counting. Work backward from the gap rather than trying to validate each conversion individually. The Conversion Quality Audit produces a daily reconciliation chart showing backend actual, Google Ads attributed, and GA4 attributed on one timeline so you see exactly where the gap sits.

Why are 70% of my PMax conversions view-through — is this real?

Probably not real in the sense of being causally attributable — PMax's channel mix typically includes 60–70% Display and YouTube impression volume, both of which generate view-through conversions aggressively when the same user later converts via another channel. A 70% view-through rate means PMax is largely claiming credit for conversions that would have happened anyway through organic search, direct, or other paid channels. The fix is to disable view-through conversions for PMax bidding (or set the view-through window to 1 hour instead of 1 day) and observe whether 'real' conversion volume holds up. The Conversion Quality Audit runs this comparison for you and quantifies the share of your PMax conversions that are click-through vs view-through.

How does Google model conversions when cookies are rejected?

When a user denies tracking cookies, Google uses three signal sources to model the conversion: (1) cross-site logged-in behaviour from users in the same cohort who did consent, (2) aggregated pattern data from advertisers with similar conversion funnels, (3) device-level and network-level signals (timing, geolocation, session characteristics) that correlate with historical conversions. The output is a probabilistic attribution that Google then reports as a conversion, with no indication in the UI that it was modeled rather than observed. In consent-mode-heavy markets this can reach 40–65% of reported conversions. The Conversion Quality Audit estimates your modeled share per campaign by comparing observed conversion rate pre- and post-consent-mode implementation.

What percentage of my conversions are modeled vs actually observed?

Google does not disclose this directly in the UI — you have to infer it. The rough method is: compare your consent-denial rate (visible in GA4's consent signal reports) against your Google Ads conversion volume, and assume Google is modeling conversions for the denied-consent user share. In the UK, consent denial averages 20–30%, so roughly 15–25% of Google Ads conversions in a UK account are modeled. In Germany that rises to 40–60%. The Conversion Quality Audit produces your specific modeled-conversion estimate and shows its impact on your effective CPA after stripping modeled conversions out.

How does poor conversion data quality affect Smart Bidding?

Smart Bidding calibrates to whatever conversion signal it receives — if 40% of your conversions are modeled, view-through, or cross-device estimates that don't represent real outcomes, the algorithm bids on traffic that 'converts' at the modeled rate but produces no actual revenue. Reported ROAS stays stable or improves while true business ROAS degrades, which is why accounts sometimes see deteriorating revenue despite 'improving' Google Ads metrics. Fixing conversion signal quality (setting high-intent events as Primary, excluding view-throughs for bidding, removing duplicates) is almost always more impactful than changing the bid strategy itself. The Conversion Quality Audit produces a Signal Health Score for each campaign showing how trustworthy its conversion data is for bidding.

Why does my Google Ads dashboard say 500 conversions but my CRM only shows 200?

A 2.5x gap is well outside the normal 15–35% discrepancy range and indicates a concrete tracking problem rather than structural difference. The most common causes: (1) a lead-gen form counting form-loads as conversions rather than submissions, (2) view-through conversions inflating the count by 2–3x (common for PMax and Display), (3) the conversion action set to 'Every' instead of 'One' for repeat form submissions, (4) duplicate conversion pixel fires. Lead-gen accounts typically see this pattern when Google Ads treats any form interaction as a conversion while CRM only records verified contacts. The Conversion Quality Audit reconciles your reported conversions against a CRM upload and identifies which of the four causes applies.

ClickCatalyst

Conversion Quality · Methodology

Back to Conversion Quality Audit

Authority Layer · 8 metrics · 7 data sources

Inside the Conversion Quality Engine

Every metric in your Conversion Quality audit is sourced from documented BigQuery queries. This page explains how we calculate modeled conversion percentages, attribution drift, deduplication risk ratios, and signal health scores — with exact thresholds for every status label.

Jump to metric

Modeled Conversion Percentage01 Attribution Drift Scatter Plot02 Conversion Type Stacked Bar03 View-Through Inflation Risk04 Deduplication Risk (Inflation Ratio)05 Conversion Lag Distribution06 GA4 Funnel vs Ads Conversions07 Signal Health Recommendations08

Data Sources

read-only OAuth · no campaign changes ever

All data is read via Google Ads API and GA4. Derived intelligence tables are pre-computed in our pipeline before your audit runs.

Google Ads API — Campaign Performance

google_ads_campaign_performance

Click, view-through, cross-device, all_conversions per campaign per day

Google Ads API — Ads Conversions

google_ads_ads_conversions

Conversion action names, categories, click vs all split, external source

Google Ads API — Keywords Conversions

google_ads_keywords_conversions

Conversion lag buckets per keyword for lag distribution analysis

Google Ads API — Hourly Pacing

google_ads_campaign_pacing_hourly

Device-level conversion quality (CVR, CPA by device)

Google Ads API — Search Terms

google_ads_search_terms_historical

Top converting terms with click-attributed conversions and CPA

GA4 — Ecommerce Funnel

ga4_ecommerce_funnel

Sessions, product views, cart adds, checkouts, purchases for funnel comparison

Derived Intelligence — Signal Health

signal_health_intelligence_daily

Pre-calculated drift, signal score, and signal status per campaign

Modeled Conversion Percentage

How much of your reported data is estimated, not observed

The share of all_conversions that are NOT click-attributed, sourced from google_ads_campaign_performance. This includes view-through, cross-device, and statistically modeled conversions. The algorithm treats every type equally when making bid decisions — it cannot distinguish hard evidence from Google's statistical estimates.

⚠️

Why this matters: A 40% modeled rate means nearly half the data the algorithm learns from is guesswork. Your ROAS and CPA numbers are inflated by these estimates. If you're making bid changes based on modeled data, you're optimizing a mirage.

💡

In plain English: We count conversions the user actually clicked for (real) and conversions Google estimated using models (estimated). The percentage tells you how much of your data is real vs guessed. Below 30% = trustworthy. Above 40% = the numbers are unreliable.

≤ 30%Clean — mostly click-attributed data

30–40%Declining — modeled data is significant

> 40%Unreliable — algorithm training on estimates

Attribution Drift Scatter Plot

Visual proof of where Ads and GA4 disagree

From signal_health_intelligence_daily: each campaign plotted with GA4 conversions (X-axis) vs Ads conversions (Y-axis). Points on the diagonal are perfectly aligned. Points above over-report in Ads. The scatter uses signal_status for color coding and campaign_name for labels.

⚠️

Why this matters: A table of drift percentages is abstract. The scatter plot makes the problem visceral — you can see which campaigns are outliers at a glance. The further a point is from the diagonal, the more the algorithm is bidding on phantom conversions for that campaign.

💡

In plain English: Each dot is one campaign. If the dot sits on the diagonal line, Ads and GA4 agree. If it's above the line, Ads is over-counting. The further from the line, the worse the tracking problem. Color tells you the severity: green = fine, amber = investigate, red = critical.

On diagonalPerfect alignment between platforms

Above lineAds over-reporting — phantom conversions

Below lineGA4 over-reporting — less dangerous

Conversion Type Stacked Bar

Click vs view-through vs modeled per campaign

From google_ads_campaign_performance: three layers per campaign bar — click_conversions (blue), view_through (amber), modeled_conversions (red). The stack visually reveals which campaigns are inflated by soft attribution. Campaigns with large non-blue segments are optimized on impression exposure rather than purchase actions.

⚠️

Why this matters: Display and Video campaigns naturally generate more view-through volume than Search. If a Search campaign has 30%+ view-through, something is wrong with tag configuration. The stacked bar catches this immediately because the colors tell the story.

💡

In plain English: Each bar is one campaign, split into three colors: blue (user clicked then bought), amber (user saw ad but didn't click), red (Google estimated). The more blue, the better. If amber or red dominate, that campaign's numbers are inflated.

Click > 70%Strong direct attribution

VT > 30%Optimizing on impressions, not clicks

Modeled > 50%Majority estimated — unreliable

View-Through Inflation Risk

Campaigns where impression exposure is counted as performance

Campaigns where view_through_conversions exceed 30% of all_conversions. VT conversions credit a campaign when a user saw (but didn't click) an ad before converting through another channel. Display and Video campaigns naturally generate more VT — Search campaigns should not.

⚠️

Why this matters: PMax and Display campaigns serve across YouTube, Gmail, and mobile apps — generating massive impression volume with low click rates. If VT conversions feed bidding strategy, the algorithm optimizes for cheap impressions rather than purchase intent.

💡

In plain English: We flag campaigns where more than 30% of conversions came from users who saw your ad but never clicked. The algorithm counts these equally with clicks — but the user never chose to interact with your ad. That's a very different signal quality.

VT < 10%Minimal inflation — mostly click-driven

VT 10–30%Some inflation — monitor

VT > 30%Impression-optimized — review bidding strategy

Deduplication Risk (Inflation Ratio)

Is the same conversion counted multiple times?

Per conversion action from google_ads_ads_conversions: all_conversions divided by click_conversions. A ratio of 1.0 means every conversion is unique. A ratio above 2.0 means each real conversion is counted at least twice — typically from duplicate tags firing in both GTM and hard-coded snippets.

⚠️

Why this matters: Duplicate conversions double the signal the algorithm uses to bid. If a purchase fires twice, the algorithm thinks the campaign converts 2x better than it does — and bids 2x more aggressively. This is the most expensive tracking bug possible.

💡

In plain English: We divide total conversions by click conversions for each tracking action. If the ratio is 2.0, each real conversion is being counted twice. Usually this means the same tag fires from two places (Google Tag Manager AND a hard-coded snippet on the page).

≤ 1.3Normal — expected VT/XD margin

1.3–2.0Suspicious — audit tag containers

> 2.0Likely duplicate tags — urgent fix

Conversion Lag Distribution

How long after a click does a conversion happen?

From google_ads_keywords_conversions grouped by conversion_lag_bucket. Shows how conversions distribute across time windows: same day, 1 day, 2–3 days, up to 30+ days. Long-tail lag means some 'waste' terms may convert later — the attribution window may be too short for your product's consideration cycle.

⚠️

Why this matters: If your attribution window is 7 days but 20% of conversions arrive after day 7, you're systematically under-counting. This makes waste calculations pessimistic and good keywords look worse than they are. High-consideration products (B2B, luxury) almost always have longer lag.

💡

In plain English: We show how long users take to buy after clicking your ad. If most buy the same day, your data is reliable. If significant purchases happen 7+ days later, any analysis using a short time window is missing conversions — and may flag good keywords as waste.

Same day > 80%Short cycle — short attribution windows are fine

7+ days > 20%Long consideration — extend attribution window to 30d

GA4 Funnel vs Ads Conversions

Where exactly does the tracking gap open?

Six funnel steps from GA4 ecommerce data (ga4_ecommerce_funnel) compared against Ads conversions: Sessions → Product Views → Add to Cart → Checkout → GA4 Purchases → Ads Conversions. If Ads Conversions exceeds GA4 Purchases, the algorithm credits conversions your analytics can't verify.

⚠️

Why this matters: This is the most intuitive visualization in the audit. A non-technical founder can see at a glance: 'My analytics says 200 purchases, but Google Ads claims 300 conversions — where did the extra 100 come from?' The answer is modeled attribution.

💡

In plain English: We build an ecommerce funnel: visitors → product views → add to cart → checkout → GA4 purchases → Ads conversions. If the Ads number is bigger than the GA4 purchases number, Google is claiming credit for conversions your own analytics doesn't see.

Ads ≤ GA4Aligned — Ads claims less than GA4 confirms

Ads > GA4Phantom gap — Ads over-claims vs analytics

Signal Health Recommendations

Prioritized fix list per campaign

Each non-HEALTHY campaign from signal_health_intelligence_daily receives a specific recommendation based on its signal_status. CRITICAL_BLIND campaigns get 'URGENT: Fix conversion tracking'. WARNING_DRIFT gets 'Audit GA4 ↔ Ads tag alignment'. LOW_VOLUME gets 'Increase conversion volume or consolidate'.

⚠️

Why this matters: Generic advice like 'fix your tracking' isn't actionable. This table gives campaign-specific recommendations ranked by severity — a prioritized task list that tells the operator exactly what to do for each campaign, in what order.

💡

In plain English: We look at every campaign that has a tracking problem and give it a specific fix instruction. The worst problems are listed first. Work top-down — each fix makes the algorithm smarter for every campaign, not just the one you fixed.

URGENTFix before any other optimization work

AUDITInvestigate tag alignment this week

CONSOLIDATEMerge low-volume into higher-volume campaigns

Proprietary notice: The methodology, scoring models, waste classification logic, and recovery projections are proprietary to ClickCatalyst Digital and provided for informational purposes only. Results are subject to standard market volatility. Recovery projections are estimates, not guarantees. Contact us within 14 days if any metric appears inaccurate.

Also worth reading

How to Read Your Conversion Quality Report →

Every section explained with exact action steps

See these metrics applied to your account

Run My Free Conversion Quality Audit