Building Analytics Infrastructure That Marketing Teams Actually Use
I've inherited broken analytics at two different B2B SaaS companies. Both times, the symptoms were the same: dashboards nobody trusted, event tracking that didn't match reality, no connection between web behavior and pipeline, and a marketing team making decisions based on vibes dressed up as data. The fix isn't installing a new tool. It's designing infrastructure that serves both marketing and engineering — and that starts with understanding why most analytics setups fail.
Why Marketing Analytics Break
The root cause is almost always the same: analytics infrastructure was bolted together reactively rather than designed intentionally. Someone added GA. Someone else configured a few events. A third person connected the CRM. Nobody documented what anything means, so when the original person leaves, the next team inherits a system they can't trust or maintain.
At Pushpay, this manifested as 600 crawl errors, misconfigured GA events, and zero ability to trace a web visitor through to a pipeline stage. The marketing team had data, technically — but nobody used it for decisions because nobody believed it was accurate. They were right not to.
The Five-Layer Architecture
When I rebuild analytics infrastructure, I think in five layers. Each one is independently valuable but designed to connect to the others.
The first layer is event taxonomy. Before you configure anything, document what you're measuring and why. Every event should have a name, a definition, a trigger condition, and a business question it answers. "Button click" is not an event taxonomy. "Demo request initiated from pricing page by enterprise segment visitor" is.
The second layer is tag management. GTM container architecture with clear naming conventions, folder organization, and server-side tagging where appropriate. The goal is a system that any competent marketer can audit and understand — not a black box that only the person who built it can navigate.
The third layer is CRM integration. Web behavior needs to connect to pipeline stages. When a visitor becomes a lead, you need to carry their web journey into the CRM so that sales has context and marketing has attribution. This is where most companies have the biggest gap — the handoff between anonymous web visitor and known contact.
The fourth layer is the data warehouse. BigQuery, Snowflake, or whatever your company uses. This is where you join web analytics with CRM data, run cross-channel attribution models, and build the dashboards that executives actually review. Without this layer, you're limited to what GA4 can show you — which is a lot, but not enough for pipeline attribution.
The fifth layer is governance. Consent management, data retention policies, right-to-deletion workflows, GDPR/CCPA compliance. This isn't optional or nice-to-have — it's a legal requirement that also happens to force good data hygiene. A consent framework that properly gates data collection is also a consent framework that ensures the data you do collect is clean and trustworthy.
What Makes This Work
Three things separate analytics infrastructure that gets used from infrastructure that gets ignored.
First, documentation. Every event, every tag, every CRM field mapping should be documented in a way that survives employee turnover. I maintain a living analytics dictionary that defines every metric the team reports on — how it's calculated, where the data comes from, and what it does and doesn't measure.
Second, auditing. I run quarterly GA4/GTM audits to catch misconfigured events, tag firing issues, and data loss. Analytics infrastructure degrades over time as websites change, new features launch, and tag configurations drift. Regular auditing catches problems before they corrupt your reporting.
Third, translation. The best analytics architecture in the world is worthless if the marketing team can't interpret it and the executive team doesn't trust it. I build dashboards that connect web metrics to the business outcomes stakeholders care about — pipeline created, revenue influenced, conversion rates by segment — not just pageviews and bounce rates.
Where to Start
If your analytics infrastructure is broken or nonexistent, resist the urge to buy a new tool. Start with an audit of what you have: document every event currently firing, check if the data matches reality (click a button, verify the event fires correctly), and identify the gaps between what you're measuring and what you need to measure.
Then design the event taxonomy before implementing anything. Get marketing, product, and engineering in a room and agree on what matters: what events, what properties, what naming conventions. This alignment meeting saves months of rework downstream.
The infrastructure should be modular — you don't need all five layers on day one. Start with taxonomy and tag management. Add CRM integration. Build toward the data warehouse. Layer in governance throughout. Each step delivers value independently while creating a foundation for the next.
The goal isn't perfect data. It's trustworthy data that enables decisions. If your marketing team is making strategy calls based on your analytics, the infrastructure is working. If they're ignoring the dashboards and going with their gut, it isn't — no matter how sophisticated the setup looks.