Skip to content
EARNST.
Tracking & Compliance

Tracking as Growth Lever: ROI, Reports & First-Party Strategy

Four tracking layers, five GA4 reports, and a first-party data strategy. The complete guide to measurable return on ad spend.

EARNST · · 40 min read

Key Takeaways

  • Smart Bidding optimizes on 60% of data: each layer recovers more. Consent +54%, SST +15 to 30%, Enhanced Conversions +5 to 15%
  • Funnel drop-off, engagement score, consent rate, attribution comparison, and new vs. returning: 5 reports that change your bids tomorrow
  • Customer Match audiences typically show 3x higher conversion rate than pixel-based lookalikes
  • Engagement score above 60 can deliver 10 to 15% lower CPA in retargeting

You have optimized your Google Ads campaigns. Keywords, bidding strategies, assets, landing pages. But the biggest lever is not in the campaign: it is underneath it.

Tracking infrastructure determines what data your campaigns optimize on. GA4 reports determine what decisions you derive from that data. And first-party data determines whether that data belongs to you or to Google.

This guide connects all three topics: from the investment calculation to the reports that actually trigger action, to the data strategy that makes you independent of platform decisions.

All percentage and EUR figures in this article are indicative values based on typical scenarios. Actual impact depends on industry, audience, existing setup, and other factors.

Table of Contents

  1. The Investment Calculation: Four Layers, One Combined Effect
  2. Google Ads Performance: How Tracking Directly Affects ROAS
  3. 5 GA4 Reports That Make Money
  4. First-Party Data Strategy: Your Data as an Asset

The Investment Calculation: Four Layers, One Combined Effect

The chain is simple: consent rate determines tracking coverage. Tracking coverage determines conversion attribution. Conversion attribution determines Smart Bidding quality. Smart Bidding quality determines ROAS.

Model Calculation at Three Ad Spend Levels

The following figures are conservative estimates for illustration. Actual impact depends on industry, audience, and existing setup.

ScenarioAd SpendData LossEffective Data BasisEstimated ROAS Loss
Smallca. EUR 5,000/month35%ca. EUR 3,25015 to 25%
Mediumca. EUR 10,000/month35%ca. EUR 6,50020 to 35%
Largeca. EUR 25,000/month35%ca. EUR 16,25025 to 40%

The loss grows disproportionately with budget. Higher budgets mean more campaign complexity. Every optimization decision is based on incomplete data, and the poor decisions compound.

Layer 1: Consent: From 55% to 85%

The consent rate is the first multiplier in the chain. Everything that happens after the banner is based on the visitors who consented.

Quick Win Checklist:

  • [ ] Measure current consent rate: GA4, Admin, Data Streams, Consent Mode Overview
  • [ ] Review banner wording: replace "Accept" with "Continue to Shop"
  • [ ] Check button hierarchy: is Accept visually dominant?
  • [ ] Check timing: does the banner appear too early? Optimal is 500 to 800ms delay
  • [ ] Reject option: visible but not prominent? Text link instead of button
  • [ ] Consent Mode v2 active? GA4, Admin, Data Streams, Web, Consent Mode

Technical Checklist:

  • [ ] Consent defaults load before GTM script (very first script block in <head>)
  • [ ] wait_for_update: 500 set (GTM waits 500ms for consent update)
  • [ ] url_passthrough: true active (click IDs are passed through)
  • [ ] ads_data_redaction: true active (PII is redacted when denied)
  • [ ] Banner dispatches consent_update custom event for GTM trigger
  • [ ] Shopify Privacy API correctly connected (setTrackingConsent)

How to build a custom banner and go from 55% to 85% is detailed in the GDPR Tracking Guide.

Layer 2: Server-Side Tracking: Recovering 15 to 30% of Users

The GTM Server Container runs on its own subdomain (e.g. analytics.yourdomain.com). DNS points via CNAME to the server. For the browser, the request looks like a first-party call: ad blockers do not filter it, ITP extends cookie lifetime to 13 months instead of 7 days.

SST Readiness Checklist:

  • [ ] GTM client-side container present and correctly configured
  • [ ] Hosting environment ready (own infrastructure or cloud)
  • [ ] DNS access to a subdomain (e.g. tracking.yourshop.com)
  • [ ] SSL certificate for the subdomain
  • [ ] GA4 server-side tag configured
  • [ ] Google Ads Conversion Linker tag in SST
  • [ ] Meta Conversions API tag in SST (with Event Match Quality above 6)
  • [ ] Effective Client-ID variable in SST (fallback to own cookie)
  • [ ] Purchase event tested: SST receives and forwards correctly
  • [ ] Fallback tested: block SST domain, standard GTM still works

Details on SST architecture and why own infrastructure beats shared hosting are in the Shopify Tracking Guide.

Layer 3: Enhanced Conversions: Better Signals for Smart Bidding

Smart Bidding receives hashed user data (email, phone, name, address) in addition to the click ID. Google can attribute conversions even when the cookie is missing or the user switches devices.

Implementation Checklist:

  • [ ] On the thank-you page: make user data from the order object available
  • [ ] SHA256 hashing server-side (in template, not in JavaScript): no PII in the browser
  • [ ] Required fields: email, phone, first name, last name, city, postal code
  • [ ] Google Ads tag in GTM: Enhanced Conversions enabled, User Data variable linked
  • [ ] In SST: Enhanced Conversions tag with server-side hashed data
  • [ ] Test: Google Ads, Conversions, Diagnostics, Enhanced Conversions Status
  • [ ] Meta in parallel: Advanced Matching with the same hashed fields

Quality Checklist:

  • [ ] Email is lowercased and trimmed before hashing
  • [ ] Phone: digits only, with country code, no spaces
  • [ ] Name: lowercase, trimmed, no titles (Dr., Prof.)
  • [ ] Hashing: SHA256, not MD5 (deprecated and insecure)
  • [ ] Purchase event and user data in the same dataLayer push (not separate)
  • [ ] first_time_accessed check: data is sent only once (no duplicate on reload)

The full implementation for Shopify is in the Shopify Tracking Guide.

Layer 4: Engagement Scoring: Feeding Smart Bidding with Purchase Intent

97% of your visitors do not buy. Smart Bidding treats them all the same: it lacks the signal for who is "hot."

Audience Strategy:

AudienceScoreBehaviorStrategy
Hot Leadsabove 60Highly engaged, did not buyRetargeting with increased bid
Warm Prospects30 to 60Moderately engagedStandard retargeting
Casual Browsersbelow 20Barely engagedLower bid or exclusion
Product ComparersAny5+ products viewed, no purchaseDynamic retargeting with bestsellers
Returning ViewersAnySame product 2+ timesUrgency messaging
Cart Abandoners (High)above 40Cart open, no checkoutHighest bid, time-limited

The full engagement scoring implementation is in the Shopify Tracking Guide.

The Combined Calculation

All figures are conservative estimates based on typical e-commerce scenarios. Actual impact varies by industry and starting point.

LayerInvestmentData GainROAS Impact
Custom CMP4 to 5 days one-time+25 to 30% consent+8 to 12%
Server-Side Trackingfrom ca. EUR 20/month + setup+15 to 30% coverage+5 to 10%
Enhanced Conversions1 to 2 days setup+5 to 15% attribution+3 to 8%
Engagement Scoring2 to 3 days setupQualitative improvement+5 to 10%
Totalapprox. 2 weeks + ca. EUR 50/month+45 to 75% data basis+20 to 40%

The effects partially multiply: more consent times more SST coverage times better attribution yields a disproportionate combined effect.

Prioritization Guide

  1. Immediately (Day 1): Measure consent rate. If below 70%: banner optimization is highest priority
  2. Week 1 to 2: Implement Consent Mode v2 correctly (possible even without custom CMP)
  3. Week 2 to 3: Activate Enhanced Conversions (quick win, low effort)
  4. Week 3 to 5: Set up SST (largest technical effort, highest long-term impact)
  5. Week 5 to 7: Implement Engagement Scoring and build audiences
  6. From Week 8: Monitoring, A/B testing of consent rate, audience performance optimization

Google Ads Performance: How Tracking Directly Affects ROAS

PMax and Its Dependency on Data Quality

Performance Max is a black box that optimizes on conversion data. The more conversions Smart Bidding sees, the more precise the bids become. At 35% data loss, PMax optimizes on a subset, and bid decisions deteriorate disproportionately.

Kill Criteria: When a Campaign Truly Does Not Work

Not every poor performance is a data problem. Sometimes the campaign genuinely does not work. Three criteria help distinguish:

CPA over 3x target after 4 to 6 weeks: The campaign has had enough learning time and still does not deliver. Check tracking setup (are all conversions visible?), then reallocate budget or pause the campaign.

Below 30 conversions per month: Not enough data for stable optimization. Options: increase budget, expand conversion actions (add micro-conversions), or switch to manual bidding.

Over 90% display share in PMax: The campaign shows almost only display ads instead of Search or Shopping. This indicates poor asset quality or too low a budget.

More on PMax learning phases and why campaigns perform poorly at the start is in the Performance Max Learning Phases article.

5 GA4 Reports That Make Money

Your GA4 has 50+ reports. Most are never opened. The few that are opened rarely deliver a clear action item. The problem is not GA4. The problem is that nobody builds the bridge between "number in the table" and "what we do differently next week."

Report 1: Funnel Drop-Off: Where Your Money Gets Stuck

This report shows how many users drop off at each step of the purchase process. From the product list to the product page, the cart, checkout, and purchase.

What decisions it enables:

  • Drop between view_item and add_to_cart over 90%: product page has a problem. Price, images, trust signals, shipping cost transparency
  • Drop between add_to_cart and begin_checkout over 70%: cart experience or shipping costs deter users
  • Drop between begin_checkout and purchase over 50%: checkout friction. Registration requirement, missing payment methods, too many steps

GA4 Setup: Explore, select Funnel Exploration. Steps: view_item, add_to_cart, begin_checkout, purchase. Breakdown by: device_category, country, traffic source. Period: at least 14 days.

Infrastructure Prerequisite: All 6 funnel events are tracked (view_item_list, select_item, view_item, add_to_cart, begin_checkout, purchase). Ecommerce clear before every push. purchase fires exactly once with correct value.

Report 2: Engagement Score Distribution: Who Is Hot and Who Is Not

This report shows the distribution of engagement scores across all visitors. What percentage are hot leads with scores above 60, how many warm prospects at 30 to 60, how many casual browsers below 20.

What decisions it enables:

  • 80% score below 20: content or UX problem. The site does not engage
  • 30% score above 60 but no purchase: conversion barrier. Price, shipping, trust, payment methods
  • Score trend is declining: traffic quality is deteriorating, check which campaigns deliver low-engagement traffic

GA4 Setup: Create custom dimension (engagement_score, event-scoped). Exploration, Free Form, dimension engagement_score, metric Users. Histogram view for score distribution.

Report 3: Consent Rate Trend: The Gatekeeper of Your Data

This report shows how your consent rate develops over weeks and months. The trend matters more than the single value.

What decisions it enables:

  • Consent rate declining: banner fatigue, UI problem, or browser update broke the banner
  • Consent rate differs by device: mobile banner needs optimization
  • Consent rate differs by country: wording or language adaptation needed

Benchmark Guide: Below 55%: action needed. 55 to 70%: average. 70 to 85%: good. 85 to 95%: excellent. Above 95%: suspicious, check whether consent is correctly captured.

Report 4: Attribution Comparison: Which Channel Actually Makes Money

This report shows how different attribution models evaluate the value of your channels. Data-Driven Attribution distributes conversion value across all touchpoints in the purchase path. Last Click gives everything to the last click.

GA4 Setup: Advertising, Attribution, Model Comparison. Compare: Data-Driven vs. Last Click. Dimension: Source/Medium or Campaign. Period: at least 30 days, ideally 90 days.

Report 5: New vs. Returning Revenue: Growth or Milking?

This report shows how much revenue comes from new customers vs. existing customers. The most important strategic question in e-commerce: are you growing, or living off repeat purchases?

What decisions it enables:

  • Over 80% new customer revenue: healthy growth, but check retention. Do they not come back?
  • Over 60% returning customer revenue: strong retention, but growth stagnating? Check acquisition channels
  • High new customer volume, low repeat: build email flows (post-purchase welcome, cross-sell, reactivation)
  • High repeat, few new customers: strengthen top of funnel. Demand Gen, Social, YouTube. Activate NCA Bidding

The Reports Together: A Dashboard That Delivers Decisions

Executive Dashboard (5 Tiles):

  1. Funnel Conversion Rate: last 30 days, trend. Where does the most get stuck?
  2. Average Engagement Score: last 30 days, trend. Is the site getting better or worse?
  3. Consent Rate: last 30 days, by device. Is the data basis growing or shrinking?
  4. ROAS by Attribution Model: Data-Driven, last 30 days. What do the campaigns really deliver?
  5. New vs. Returning Revenue Split: last 30 days. Growth or milking?

Meeting Cadence: Weekly: funnel drop-offs and consent rate. Monthly: attribution comparison and revenue split. Quarterly: engagement score trend and audience performance.

First-Party Data Strategy: Your Data as an Asset

Your Google Ads campaigns generate thousands of data points every month: who clicks, who buys, who returns. But this data does not belong to you. It belongs to Google. And Google uses it: for you and for your competitors.

The analogy is simple: you invest in a house, but the land belongs to someone else. First-party data is the land.

The Data Pyramid

Zero-Party Data: Directly communicated by the customer. Preferences, surveys, quiz answers, wishlists. Highest quality but hard to scale.

First-Party Data: Collected by you on your platforms. Website behavior, shop interactions, email engagement, CRM data. You control collection, storage, and use. This is the core of this section.

Second-Party Data: Shared by partners. Relevant for large enterprises with partnership ecosystems, less so for SMEs.

Third-Party Data: Bought from data brokers. Cookies and tracking pixels that collect across websites. Dying out: technically through cookie deprecation, legally through GDPR.

What Is Not First-Party Data

The GA4 Client ID belongs to Google. Safari deletes it after 7 days. Google Ads conversion data is stored at Google. You see reports, but the raw data is not yours. Facebook Pixel data is stored at Meta. All three are rentals, not ownership.

Why Third-Party Is Dying

In 2020, Safari ITP blocked third-party cookies completely. In 2023, Firefox ETP followed. In 2026, the reality is: third-party cookies are already dead for 40 to 50% of users. Not theoretically, not planned: now.

Stage 1: Own Visitor Identity (Foundation)

Your own visitor identity is insurance against all browser updates and platform changes. If Safari deletes the _ga cookie after 7 days, you still have your own ID.

The technical implementation: one UUID per visitor, stored in a first-party cookie. Server-side cookie setting via the SST container gives the cookie a lifetime of 13 months, even with Safari. The GA4 Client ID is stored as backup. At login, the visitor ID links to the customer ID: deterministic cross-device identity.

Stage 1 Implementation Checklist:

  • [ ] Define own visitor ID cookie (name, structure, domain)
  • [ ] Implement UUID generation in tracking JavaScript
  • [ ] Cookie is set on first visit (first-party, Secure, SameSite=Lax)
  • [ ] Visit count is incremented on each visit
  • [ ] GA4 Client ID extracted from _ga cookie and stored in own cookie
  • [ ] SST sets the cookie server-side (HttpOnly, 13-month lifetime)
  • [ ] dataLayer contains visitor_id, visit_count, is_returning on every pageload
  • [ ] At login: customer_id is linked with visitor_id

Stage 2: Engagement and Behavioral Data (Enrichment)

Not just who was there, but how engaged, and across sessions. A visitor who scrolls through 5 product images, spends 3 minutes on the page, and opens a product description has different purchase intent than someone who bounces after 10 seconds.

The engagement score (0 to 100) per session quantifies this difference. Cross-session product interest stores which products a visitor viewed across multiple visits. Returning product view flags detect when someone views the same product a second time.

Stage 2 Data Inventory:

SignalStorageDurationUsage
Engagement ScoredataLayer, GA4SessionAudiences, Bidding
Products Viewed (Session)sessionStorageSessionFunnel Analysis
Products Viewed (Lifetime)localStoragePersistentCross-Session Retargeting
Returning Product ViewlocalStoragePersistentUrgency Messaging
Cart AbandonmentdataLayer, GA4EventEmail Trigger, Retargeting
Scroll DepthdataLayer, GA4EventContent Optimization
Active TimedataLayer, GA4EventEngagement Segmentation

Stage 3: CRM Integration and Predictive Audiences (Scaling)

The third stage connects anonymous website behavior with known customer data and creates competitive advantages that last.

Customer Match: Upload CRM emails as audiences to Google Ads and Meta. Google and Meta match hashed emails with their own user data and create high-quality lookalike audiences. Quality: significantly better than pixel-based lookalikes because the seed data comes from actual buyers.

Predictive Audiences in GA4: "Likely to purchase in 7 days." GA4 automatically creates audiences based on machine learning. Prerequisite: sufficient purchase volume (over 1,000 purchases in 28 days). For shops with less volume: the manual score-based audiences from Stage 2 are the alternative.

CLV-based bidding strategy: High-CLV customers get higher bids. A customer with ca. EUR 2,000 annual revenue justifies the acquisition investment that a one-time buyer at ca. EUR 30 does not.

Email segmentation by engagement score: Hot leads (score above 60) get different email flows than casual browsers. Higher relevance, higher open rates, more revenue from owned media.

RFM Analysis (Recency, Frequency, Monetary): Automated customer segments based on purchase behavior. VIP customers, churn risks, new customers with potential.

Stage 3 Readiness Assessment:

  • [ ] CRM or email tool available (Klaviyo, Mailchimp, ActiveCampaign, etc.)
  • [ ] Customer list with over 1,000 email entries
  • [ ] GA4 User-ID matching activated (at login)
  • [ ] Google Ads Customer Match set up
  • [ ] Sufficient purchase volume for GA4 Predictive Audiences (over 1,000 purchases per 28 days)
  • [ ] CLV calculable per customer (from Shopify or CRM)

The Privacy Framework

First-party data and GDPR are not contradictory. Your own data means more control and therefore better compliance.

What you may do with first-party data (with consent):

  • Set own cookies
  • Collect anonymous engagement data (scroll depth, active time, without PII)
  • Use hashed email addresses for Customer Match
  • Cross-session tracking on your own domain
  • Server-side cookie setting for longer lifetimes (13 months)
  • Use CRM data for segmentation and personalization

What you must not do:

  • Store or transmit PII in plain text (always SHA256 hash)
  • Collect data without consent (Consent Mode defaults must be "denied")
  • Share data with third parties without legal basis
  • Track users across domains without explicit consent

The full legal framework is in the GDPR Tracking Guide.

The 90-Day Plan: From Zero to Your Own Data Strategy

Day 1 to 30: Lay the Foundation

  • [ ] Implement Consent Mode v2 correctly (or custom CMP)
  • [ ] Bring consent rate above 75%
  • [ ] Implement own visitor identity (own cookie, UUID, visit count)
  • [ ] Set up SST (if not already in place)
  • [ ] Activate Enhanced Conversions (Google and Meta)
  • Milestone: Tracking coverage above 80%

Day 30 to 60: Enrich Data

  • [ ] Implement engagement scoring (scroll, time, interactions)
  • [ ] Activate cross-session product interest (localStorage)
  • [ ] Set up cart abandonment signal
  • [ ] Create GA4 custom dimensions for all signals
  • [ ] Define first audiences in GA4 (hot leads, cart abandoners, product comparers)
  • Milestone: 5+ active audiences in GA4

Day 60 to 90: Connect and Scale

  • [ ] Import GA4 audiences into Google Ads
  • [ ] Differentiate bidding by audience segments
  • [ ] Set up Customer Match with CRM emails
  • [ ] Segment email flows by engagement score
  • [ ] Set up CLV calculation from Shopify or CRM
  • [ ] First performance analysis: before/after ROAS comparison
  • Milestone: Measurable ROAS improvement, own data foundation growing

ROI of a First-Party Data Strategy

Short-term (Month 1 to 3): 10 to 20% more attributed conversions through 13-month attribution. 10 to 15% lower CPA on retargeting through hot-lead focus. Setup cost typically pays back in 2 to 3 months.

Medium-term (Month 3 to 12): Cross-session data reaches critical mass. Customer Match audiences deliver higher-intent lookalikes: typically 3x better prospecting performance. Email segmentation by engagement increases open and click rates. ROAS uplift 10 to 25%.

Long-term (12+ months): Own data foundation as an asset. 50,000 to 200,000 identified visitors with engagement history. 5,000 to 20,000 CRM profiles with CLV data. Portable data foundation when switching platforms. Competitors without their own data pay 20 to 40% higher CPAs.

Conclusion

Tracking infrastructure is not an IT expense. It is the most profitable investment in your marketing stack. Four layers can deliver 20 to 40% more ROAS. Five GA4 reports replace 50+ unused standard reports and deliver decisions instead of numbers. And a first-party data strategy makes you independent of platform decisions, browser updates, and cookie deprecation.

You now know what to do and in what order. If you want to check the current state of your setup, start with our Tracking Audit. And if you do not want to implement it yourself: we do it for you.

The 4 Tracking Layers and Their Cumulative Impact

ROI of Tracking Infrastructure Over Time

Costs (Setup + Operation) Savings through Better Data

Our service

Tracking & Data Architecture

20–40% of your conversion data is missing. Server-side tracking, Consent Mode v2, 18+ events, and engagement scoring bring it back.

Learn more

How can we help you?

Choose your perspective — we'll tailor the content for you.