Marketing Intelligence Across 60+ Websites
Problem
When a company runs dozens of websites, performance data is normally collected where it is produced. Each website has its own Search Console property and its own analytics setup, and every team reads the numbers in the way that suits its own site. This works while the sites are few and each is judged separately.
The contradiction is that the sites are managed as one portfolio, yet their performance is stored and defined differently. Without a shared data model, teams cannot reliably compare performance between websites, calculate return on investment, attribute results across channels, or build consistent reporting. A new dashboard would not change that, because a dashboard only displays whatever the data underneath it already contains. What does a comparison between two websites mean when each one measures success differently? And how can a company decide where to invest if its own numbers cannot be compared?
Action
The better principle is to settle the data model before building any report. I designed the foundation schema for collecting Google Search Console and analytics data from more than 60 websites and centralised it in a BigQuery data warehouse.
The system was designed to support real-time performance tracking, ROI calculations, customisable reporting, multi-channel attribution, campaign optimisation and cross-site analysis. I also structured the work around collaboration between the SEO, content, marketing and project-management teams, so that reporting could support decisions instead of existing as an isolated analytics layer.
Result
A fragmented reporting environment became a centralised marketing intelligence system, capable of comparing and analysing performance across more than 60 web properties. The verified proof is:
- 60+ websites integrated
- a centralised BigQuery warehouse
- a shared analytics schema
- a cross-functional reporting foundation
The lesson
Reporting problems are often data-model problems in disguise. If every property defines and stores performance in its own way, the first step is to agree on what is being measured, and only after that does it make sense to decide how it should be displayed.