DatascaleBlog
Blog
Field notes, not a content pipeline.
We write when we have learned something that saves someone else work. No editorial calendar, no SEO filler.
25 posts
Google Tag Manager needs consent: the VG Hannover ruling
A German court has ruled that loading gtm.js already accesses the user's device and requires prior consent under Section 25(1) TDDDG.
stape.io Reviewed: Managed sGTM Hosting, With the Maths
stape.io as managed hosting for server-side GTM: plans from $0 to $167 a month, billing by request, EU region. With the cost comparison against Cloud Run, the four scoring axes from our catalogue, and our partner disclosure.
JENTIS Reviewed: What Twin Server and Synthetic Users Actually Deliver
JENTIS in 2026 as a server-side platform out of Vienna: twin-server architecture, field-level data reduction, ISO 27001. With an honest read on Synthetic Users, the pricing question, and the comparison against stape.io and your own sGTM.
Fivetran Reviewed: What the dbt Merger Changes for Marketing Data
Fivetran in 2026 as a managed ELT platform: MAR billing with no published plan prices, 750+ sources, and a completed merger with dbt Labs as of 1 June 2026. With the four scoring axes from our catalogue and the comparison against funnel.io and Airbyte.
Consent Mode, four weeks in: three findings from DACH audits
Four weeks after the June 15, 2026 change: Google Signals is gone as a fallback, silent consent gaps are emptying remarketing lists, and Swiss traffic gets its own CMP rules.
Server-Side Tracking for Shopify with Stape: The Complete Guide (and the Sandbox Traps Every Other Guide Misses)
Server-side tracking for Shopify with Stape, from scratch. The sandbox gotchas every other guide misses: iframe URLs in GA4, lost add-to-carts, broken attribution.
The EU AI Act Just Slipped: What the Digital Omnibus Means for Marketing Teams
The Digital Omnibus defers the EU AI Act's high-risk obligations to December 2027. What stays, what moves, and why it's no reason to park your AI inventory.
Usercentrics + Server-Side GTM: Wiring the Consent Chain Correctly
Wiring Usercentrics into server-side GTM: load the default state before GTM, enable the Google Consent Mode integration, verify consent signals reach the server container. With a QA matrix and the 5 most common audit findings.
Why Most BigQuery Projects Fail (and It's Almost Never BigQuery)
BigQuery projects rarely fail on the tool. They fail on raw GA4 export with no model, SELECT * eating the budget, and missing tests. Four patterns, one counter-design.
Composable CDP vs. Packaged CDP: The Bitter Truth for Mid-Market
Composable CDP sounds like progress in 2026, but in mid-market it often fails on hidden engineering costs. When packaged wins, when composable wins, honestly decided.
Dashboards that get used. 6 design patterns from 5 years of practice
Most marketing dashboards get built, presented once, then never opened again. Six patterns that make the difference between decoration and decision tool, even in the age of AI-generated insights.
Data Studio vs Power BI vs Tableau 2026: Which Tool for Which Marketing Team?
Three BI tools, three very different strengths, and in 2026 three very different AI layers (Gemini, Copilot, Einstein/Pulse). Which one fits your team answers the question better than any marketing pitch.
What Marketing Analytics Has to Learn from the EU AI Act (Even If You Don't Build AI)
As of August 2026: the Article 50 transparency obligations apply, the high-risk obligations only from December 2027. Which marketing setups count as AI systems, which risk tier they fall into, and what to do about it now.
Funnel.io Review 2026: Where It Sits in the Modern Data Stack, and What It Costs
Funnel.io evaluated as an ELT tool for the modern data stack: built-in harmonisation before BigQuery, dbt-compatible architecture, AI-ready output. With pricing estimator, dbt-vs-Funnel-UI comparison, and head-to-head vs Supermetrics and Fivetran.
Server-Side Tracking with stape.io 2026: When It's Worth It, and When It Isn't
Server-side tracking is the 2026 default, third-party cookies are dead, Safari ITP wipes client-side cookies after 7 days. The sGTM container doubles as a data firewall in front of Meta CAPI, Google, and TikTok. Honest take on Stape vs GCP.
Firebase Analytics and GDPR 2026: What Applies to App Data
Firebase Analytics in 2026: IP anonymisation is the default, the real mandatory list is EU data routing, Consent Mode v2 (DMA-enforced), and the iOS dual-consent ATT + GDPR. Plus the Art. 9 special case for health apps.
OneTrust vs Usercentrics vs Cookiebot 2026. Who Each One Fits (and When All Three Are Wrong)
OneTrust, Usercentrics, and Cookiebot solve the same problem at three complexity levels. In 2026 a fourth option joins: the headless CMP. Includes Cookiebot's pricing trap, SPA pitfalls, and an interactive feature matrix.
The Measurement Blueprint 2026: Spec, Data Contract, and AI Readiness in One
In 2026 a measurement blueprint is more than an Excel doc: a versioned JSON schema in a git repo that blocks bad tracking before it can merge, and the foundation for trustworthy BI, CDP, and AI data.
Plausible Analytics for Mid-Market 2026: When It Beats GA4, and When It Doesn't
Plausible in 2026: 100% deterministic data without a cookie banner, against GA4's 30–50% AI-modelled estimates under Consent Mode v2. Honest take with an interactive decision engine.
Consent Mode V2 in Practice 2026: What It Can, What It Can't. Why 80% Are Broken
Consent Mode V2 in 2026: gcd parameter for debugging, basic mode as the lower-risk default (advanced mode under EU DPA pressure), and the 5 SPA race conditions where 80% of setups stumble.
GA4 Audit: 7 Errors We Find in Almost Every Setup, and What They Cost
Three years of GA4 audit findings: the seven errors that show up so consistently we now check for them by default. Sorted by risk, with concrete diagnosis steps and business impact per error.
Cookieless Attribution 2026: What Actually Works, and What Is Marketing Hype
Third-party cookies are dead in 2026. No single method is enough, the 2026 answer is triangulation: server-side tracking + MMM + incrementality testing, calibrated against ground truth.
What Is a dataLayer? The Heart of Marketing Tracking 2026
The dataLayer is more than pixel food: it's the single source of truth for server-side tracking, Consent Mode v2 state management, and data contracts. Without it AI/BI models hallucinate; with it, data quality stays consistent across platforms.
Benefits of Server-Side Tracking: Why It's Mandatory in 2026
Performance, data quality and GDPR governance compared head-to-head: why server-side tracking is the new baseline for serious tracking in 2026, and why client-side setups hit structural limits.
No post in this category.
From practice
What you read here, we test against your setup.
The Audit Sprint turns the principles in these articles into a ranked list of findings for your architecture.