Marketing Data Stack Architecture: A Reference Blueprint by Company Size

Why Architecture Matters Before Tool Selection

The most common data stack mistake: selecting tools before designing the architecture. The result is a collection of disconnected tools with overlapping coverage and gaps in between — expensive to maintain and impossible to analyze across.

Start with architecture (what data flows where and why), then select tools that fit the architecture. Our marketing analytics team designs data stacks aligned to company size and current needs — not maximum sophistication.

Startup Stack (0–50K monthly sessions, <£1M revenue)

Priority: Low cost, fast setup, avoid over-engineering

Web tracking: GA4 (free) + GTM (free)
Ad platforms: Google Ads, Meta Ads (native reporting)
Email: Mailchimp or Klaviyo (built-in analytics)
CRM: HubSpot Free or Pipedrive
Reporting: GA4 built-in + Looker Studio (free)

Data flows: GA4 receives all web events via GTM. Ad platforms integrate natively with GA4 for conversion import. Looker Studio connects to GA4 and ad platforms for consolidated reporting.

What you get: Web behavior, channel attribution, conversion tracking, basic email analytics, sales pipeline data.

What you can't do yet: Cross-channel customer-level analysis, product analytics, offline data integration.

Monthly cost: £0–£200 (tool licenses only; GA4 and GTM are free)

Growth Stack (50K–500K monthly sessions, £1M–£10M revenue)

Priority: Customer data unification, better attribution, warehouse analytics

Web/app tracking: GA4 + GTM + Server-side events
CDP: Segment or RudderStack Cloud
Data Warehouse: BigQuery (GA4 export + CDP events)
Transformation: dbt (optional at this stage)
Email/CRM: Klaviyo, HubSpot, or Salesforce
Ad platforms: Google Ads, Meta, LinkedIn
BI/Reporting: Looker Studio + BigQuery

Key additions from startup:

• Server-side tracking for better data quality and privacy compliance
• CDP for identity resolution and audience activation
• BigQuery for cross-source analysis

Monthly cost: £1,500–£5,000

Scale/Enterprise Stack (>500K sessions, >£10M revenue)

Priority: Full customer journey visibility, ML, organizational data governance

Collection: GTM Server-Side + CDP (Segment/mParticle)
Warehouse: BigQuery or Snowflake
Transform: dbt Core + CI/CD
Reverse ETL: Hightouch or Census
BI: Looker or Tableau
ML: Vertex AI or SageMaker (LTV prediction, churn)
Experimentation: Optimizely or Statsig
Data governance: Data catalog + access controls

Monthly cost: £15,000–£100,000+

Common Architecture Mistakes

• Building enterprise architecture at startup stage: Adds cost and complexity before you have the data volume to benefit
• No warehouse at growth stage: Once you have multiple channels and need cross-source analysis, warehouse is essential — delaying creates technical debt
• Tool-first decisions: Buying a CDP before defining what audiences you'll activate creates "CDP trophy" syndrome — expensive tool, no use case

Our analytics consulting team designs data stack architecture appropriate for your current and near-future needs. Contact us for a data stack architecture review.