From Data Chaos to Triple ROAS
How an established e-commerce company made its ad spend significantly more profitable through server-side tracking (measured ROAS increase: approx. 3×) — while reducing data loss from 42% to under 5%.
Background & Challenge
An established e-commerce company with seven-figure annual revenue was fighting an increasingly severe data loss problem: Safari ITP, iOS ATT and widespread ad blockers were causing 42% of all tracking events to disappear — without GA4 giving any indication. The result was campaign decisions made on incomplete data and a ROAS that was structurally reported too low.
- 42% data loss due to ad blockers and Safari ITP
- Over 60 outdated tags in the GTM container, some firing before the consent signal
- No first-party data strategy — all cookies set as third-party
- Meta CAPI not implemented — conversions were massively under-reported
- GA4 and Google Ads showed >30% divergence on conversion figures
Our Approach
Analysis & Audit (Week 1–2)
Complete inventory of the existing GTM container, all tags and tracking endpoints. Quantification of data loss from ad blockers and ITP across all browsers and devices.
Server-Side Infrastructure (Week 3–4)
Setup of the GTM Server Container on Google Cloud. Configuration of the first-party cookie domain and SSL certificate for reliable data capture.
GA4 & Meta CAPI Migration (Week 5–6)
Step-by-step migration of all events to the server container. Parallel operation to validate data quality before cutting over completely.
Consent Mode v2 & Validation (Week 7–8)
Clean integration of CMP signals into the server layer. QA with DebugView, GA4 Realtime and Meta Test Events to confirm full compliance and accuracy.
Go-Live & Handover (Week 9–10)
Production cutover, deactivation of old client-side tags, documentation and handover to the internal team with full training.
Key Learnings
- Server-side tracking is not a luxury for e-commerce above CHF 1M revenue — it is a necessity
- The biggest lever is often not a new tool but improving the data quality of existing ones
- Parallel operation (client + server) is essential for a safe migration without data gaps
- Consent Mode v2 must be initialised before all GTM tags — sequence is everything
- GTM audits regularly uncover privacy risks that no one had on their radar
Making the Lead Pipeline Visible
A Swiss B2B SaaS provider didn't know which marketing channels were truly bringing customers. After 8 weeks they had a complete picture of their lead journey — and halved their CPL.
Background & Challenge
A Swiss B2B SaaS provider with a growing inside sales team faced a classic problem: marketing was investing budget in SEO, Google Ads, LinkedIn and content — but which channel was actually bringing paying customers remained unclear. Three different tools produced three different numbers. Budget allocation decisions were made on gut feeling, not reliable data.
- Three different analytics tools producing contradictory conversion figures
- Form events only partially captured — up to 40% of demo requests went untracked
- No CRM integration: offline conversions and opportunity data never fed into analysis
- LinkedIn Ads and Google Ads had separate attribution — no cross-channel view
- CPL optimisation was impossible without reliable lead source attribution
Our Approach
Analytics Audit & Data Inventory (Week 1–2)
Full analysis of all existing tracking implementations, form events and conversion goals. Identification of data gaps per channel and tool.
GA4 Rebuild & Complete Form Tracking (Week 3–4)
Clean GA4 implementation with comprehensive capture of all lead forms, demo requests and trial sign-ups via a structured GTM data layer.
CRM Integration & Offline Conversions (Week 5–6)
HubSpot CRM connected via API. Lead status and deal stage fed back into GA4 and Google Ads as offline conversions — enabling true end-to-end attribution.
Multi-Channel Attribution & Data Studio Dashboard (Week 7)
Unified reporting across all channels in a single Data Studio dashboard — from first impression to closed deal, including CPL trends per channel.
Team Handover & Training (Week 8)
Documentation, handover to the marketing team and training of internal users for independent work with the dashboard and new reporting structures.
Key Learnings
- CRM integration is the decisive step: only when offline data flows back does a complete attribution picture emerge
- Cross-channel attribution starts with clean single-source tracking — not an expensive attribution tool
- Consistent form tracking in GA4 is what makes real conversion optimisation possible
- B2B SaaS benefits disproportionately from Smart Bidding once offline conversions are correctly passed through
- A single dashboard used by all stakeholders is more valuable than ten specialised views
Scalable Tracking Infrastructure for an Agency
How a Swiss performance agency standardised and automated its tracking quality across 12 client accounts — with a white-label tracking stack that deploys in hours rather than weeks.
Background & Challenge
A Swiss performance agency was managing 12 client accounts simultaneously — with steadily growing coordination overhead. Every new client required manual GTM configuration, server-side tracking setup from scratch and a fresh Consent Mode integration. That consumed 3–4 weeks per onboarding, tied up resources and produced inconsistent data quality across accounts. At the same time, there was no central monitoring: tracking errors often went unnoticed for weeks.
- 3–4 weeks of effort for tracking onboarding of every new client
- No standardised GTM container — every implementation was individual and hard to maintain
- No central monitoring infrastructure: errors in one account often went undetected for weeks
- Inconsistent Consent Mode implementations increased GDPR risk across all client accounts
- Team overloaded by manual maintenance work: 40% of capacity absorbed by legacy upkeep
Our Approach
Infrastructure Analysis & Template Design (Week 1–3)
Inventory of all 12 accounts. Derivation of shared requirements. Design of a white-label GTM template container with standardised tags, triggers, variables and consent logic.
Server-Side Tracking Platform Build (Week 4–6)
Setup of a central GTM Server Container infrastructure on Google Cloud with multi-tenant routing logic — one infrastructure serving all current and future client accounts.
Template Rollout Across All 12 Accounts (Week 7–9)
Step-by-step migration of all existing accounts to the template container. Parallel operation for validation without data loss. Per-account customisations remain possible.
Automated QA & Account Monitoring (Week 10–11)
Automated checks via GA4 Data API: alerts for traffic anomalies or conversion drops across all 12 accounts simultaneously — proactive rather than reactive.
Documentation & Team Enablement (Week 12)
Complete technical documentation, a client onboarding playbook and training of the agency team for independent use, development and future client onboarding.
Key Learnings
- Standardisation is the biggest lever for agencies: one template container beats 12 individual solutions in both efficiency and quality
- Central server-side tracking reduces not only costs but also privacy risks across all clients
- Automated monitoring is the difference between reactive error correction and proactive quality assurance
- Onboarding playbooks pay for themselves after just the second new client
- Multi-tenant infrastructure is economically superior from 5 parallel accounts onwards
Consent Mode v2: Privacy Without Data Loss
How a Swiss media company increased its analytics coverage from 58% to over 95% through correct Consent Mode v2 implementation — without compromising nDSG compliance.
Background & Challenge
An established Swiss media company had implemented a CMP following the introduction of the nDSG, considering their data privacy obligations covered — or so they thought. What nobody had noticed: Consent Mode v2 had never been correctly integrated into GA4 and GTM. The result was a silent data loss of around 40%, which triggered no alerts and no error messages. Campaign decisions, content analyses, and reach measurements were based on structurally incomplete data for months.
- ~40% data loss due to missing Consent Mode v2 signals
- GA4 received no consent_status parameters from the CMP
- GTM tags fired before the CMP callback — incorrect firing order
- Modelled data in GA4 not activated due to missing prerequisites
- GA4 and Google Ads showed >25% divergence in conversion figures
Our Approach
Audit & Diagnosis (Week 1)
Full analysis of the GTM container and CMP setup. Identification of the incorrect tag firing order and missing dataLayer integration between Usercentrics and GTM.
Consent Mode v2 Implementation (Week 2)
Correct integration of default and update commands via GTM. Ensuring the CMP callback executes before all tracking tags. Activation of Advanced Consent Mode in GA4.
Validation & Go-Live (Week 3)
QA using GTM Preview, GA4 DebugView, and the Consent Mode Debugger. Review of modelled conversions in GA4. Documentation and handover to the internal team.
Key Learnings
- Consent Mode v2 is not optional — mandatory since March 2024 for Google Ads conversion modelling
- A CMP alone is not enough: the technical integration into GTM and GA4 is what matters
- Tag firing order in GTM is the most common and most invisible misconfiguration
- Advanced Consent Mode enables GA4 modelling — but only with correct signal transmission
- Data loss from incorrect consent setup never appears as an error — it shows up as a silent gap
Case Study 05 — Offline and Online — Finally Speaking the Same Language
How a Swiss multichannel retailer linked its in-store sales to online campaigns for the first time — discovering that its supposedly weakest channel was actually its most profitable.
Background & Challenge
A Swiss multichannel retailer with seven stores and a growing online shop faced a classic blind spot: over 60% of its revenue came from in-store — yet not a single one of these purchases fed into marketing attribution. Google Ads was optimising on online conversions and seeing only a fraction of the true impact. The result: a channel producing strong offline results was continuously underbudgeted — because the numbers simply didn't show it.
- Over 60% of purchases happened in-store — entirely outside the analytics attribution
- CRM data (loyalty cards, POS transactions) was not linked to online touchpoints
- Google Ads optimised exclusively on online conversions — offline ROAS was unknown
- Different stores used different POS systems with no unified data interface
- Customer journeys spanned multiple channels and days — with no connecting ID
Our Approach
Data Audit & Matching Concept (Weeks 1–2)
Analysis of all available data sources: POS systems, CRM, online shop and GA4. Development of a privacy-compliant matching concept based on hashed email addresses (GCLID matching).
Offline Conversion Import Setup (Weeks 3–4)
Technical implementation of the Google Ads Offline Conversion Import. Automated export of POS transactions via BigQuery. Daily upload of hashed customer data as offline conversions.
Enhanced Conversions & First-Party Data (Weeks 5–6)
Implementation of Enhanced Conversions for Web to improve online attribution. Clean integration of the first-party data flow from the online shop into GA4 and Google Ads.
Cross-Channel Dashboard (Week 7)
Development of a BigQuery-powered reporting dashboard consolidating online and offline ROAS per campaign and channel — for the first time in a single view.
Validation & Go-Live (Week 8)
QA of the entire conversion chain from first click to in-store purchase. Training of the marketing team and handover of the reporting infrastructure.
Key Learnings
- Offline attribution gaps systematically lead to underinvestment in the most profitable channels
- Enhanced Conversions are the technical prerequisite for valid cross-channel attribution
- Privacy-compliant offline matching works without compromise — hashed IDs are sufficient
- BigQuery as a data hub is the key enabler for complex multichannel attribution
- Only when offline data feeds into Smart Bidding can Google Ads reach its full potential
Case Study 06 — Recruiting Tracking Without Compliance Risk
How a Swiss industrial company tracked its recruiting campaigns fully for the first time — discovering that LinkedIn was half as expensive as assumed, but twice as effective.
Background & Challenge
A Swiss industrial company running ongoing employer branding campaigns on LinkedIn, Meta and Google was investing significant budget in recruiting ads — without knowing which platform was actually generating applications. The ATS provided application numbers, but no channel attribution. At the same time, the company was required under the nDSG to track in a privacy-compliant manner — a CMP was implemented, but the technical foundation was missing.
- No attribution of applications to marketing channels — LinkedIn, Meta and Google all generated clicks, but who actually converted was unknown
- Consent Mode v2 not correctly implemented — around 35% of events disappeared after CMP rejection
- LinkedIn Insight Tag fired without consent check — GDPR/nDSG risk
- ATS system (applicant tracking) generated no GA4 events on form completion
- Meta campaigns running without CAPI — conversions massively underreported
Our Approach
Audit & Compliance Check (Week 1)
Full analysis of all active tracking implementations, CMP integration and consent flows. Identification of GDPR/nDSG gaps and missing events.
Consent-Compliant Event Setup (Weeks 2–3)
Clean re-implementation of all pixels (LinkedIn, Meta, Google) with correct Consent Mode v2 integration. All tags fire exclusively after a positive consent signal.
ATS Integration & Conversion Tracking (Weeks 3–4)
Technical integration of the ATS form via GTM. Full tracking of the application journey: job listing → click → form start → submission — with channel attribution in GA4.
LinkedIn & Meta CAPI Implementation (Week 5)
Server-side delivery of application conversions to LinkedIn and Meta via Conversion API — for correct attribution even when consent is declined.
Reporting Dashboard & Handover (Week 6)
Setup of a reporting dashboard with CPAp (cost per application) per channel, quality metrics from the ATS and monthly trends. Handover to HR and marketing teams.
Key Learnings
- Recruiting tracking is technically identical to e-commerce attribution — the conversion is just called "application" rather than "purchase"
- Consent Mode v2 is mandatory for HR campaigns too — LinkedIn and Meta have their own requirements
- Server-side CAPI safeguards attribution signals even at high consent rejection rates in HR environments
- A correctly measured cost per application makes budget decisions between channels possible for the first time
- ATS integration pays off with as few as 50 applications per month
Case Study 07 — GA4 Compliance in Regulated Finance
How a Swiss financial services firm introduced privacy-compliant analytics without US data transfer — gaining reliable user data for digital channel optimisation for the very first time.
Background & Challenge
A regulated Swiss financial services firm faced a fundamental conflict: marketing needed reliable user data to optimise digital campaigns, while compliance banned any transfer of personal user data to US-based servers — a prohibition that carried more weight than ever following the Schrems II ruling and the nDSG. Standard GA4 tracking was therefore simply not an option.
- Compliance requirement: no raw data on US servers — standard GA4 setup not permissible
- Previous tracking abstinence had created a complete blind spot in digital marketing
- No Consent Mode v2 implemented — despite active Google Ads campaigns
- Sensitive page views (product pages, account opening flow) could not be passed on unfiltered
- Internal IT requirement: no third-party cookies, no client-side fingerprinting
Our Approach
Compliance Mapping & Technical Concept (Weeks 1–2)
Close collaboration with compliance and legal teams. Documentation of all data requirements and restrictions. Development of a technical architecture enabling analytics while meeting all regulatory requirements.
Build the Server-Side Proxy Layer (Weeks 3–5)
Setup of a GTM Server Container in a Swiss Google Cloud region (europe-west6). All GA4 events are routed through a proprietary proxy endpoint — IP addresses are stripped before forwarding, no personal raw data reaches Google's servers.
Data Masking & Event Filtering (Weeks 6–7)
Implementation of server-side filtering logic: sensitive page views are anonymised, user IDs are hashed, no names or account numbers appear in events. Only aggregated, non-personal events flow into GA4.
Consent Mode v2 & Google Ads Integration (Weeks 8–9)
Correct consent integration for Google Ads conversion modelling. As no client-side tracking is possible, all conversions run through the server layer with complete consent signal forwarding.
Compliance Documentation & Go-Live (Week 10)
Full technical documentation of the data flow for the compliance team. Data Protection Impact Assessment (DPIA) supported. Go-live after internal sign-off by compliance and IT.
Key Learnings
- Server-side tracking is the only route to compliance-grade analytics in regulated industries
- IP stripping and server-side event filtering are technically trivial — but often organisationally unresolved
- Compliance teams are partners, not blockers — early involvement saves weeks
- EU cloud regions (europe-west6) don't fully resolve the Schrems II issue, but significantly reduce risk
- Without a tracking foundation, data-driven marketing in regulated industries is simply not possible