Case study · Hub
Content & Data Hub: AI-Powered Data, Content & Automation Platform
One platform connecting performance data, content generation and automated testing across 12 countries and 9 languages, for content, SEO, growth and product teams.
- Business area
- Content and growth operations
- Users
- Content, SEO, Growth and Product
- Technology
- Next.js, TypeScript, a rules engine, live search
- Outcome
- Centralised decisions, content, optimisations and learning
The problem
What was getting in the way.
Content and SEO teams manage comparison content across dozens of business verticals in 12 countries and 9 languages, while performance data sat separately in analytics, BI and experimentation tools. Nothing connected what changed in performance to what content should happen next, or which tests were worth running. Teams were left jumping between dashboards, guessing at causes, and losing hard-won learnings over time.
Why it was hard
Scale
Dozens of verticals across 12 countries and 9 languages, each with a different retail calendar.
Disconnected systems
Performance data, content planning and experiment results lived in separate tools.
No causal insight
Dashboards showed what changed, but never explained why, or what to do next.
Lost learnings
Test results and lessons disappeared into dashboards and documents instead of compounding.
The solution
What was built.
Designed Content Hub as a three-layer platform: a data intelligence layer that diagnoses why performance moves and ranks likely causes, a content layer that matches seasonal and commercial events to the right verticals and generates localised copy in 9 languages, and an automation layer that turns diagnoses into structured experiments and feeds outcomes back into a compounding learning memory.
Design decisions
- Treat content generation as one layer of a bigger data-to-decision system, not a standalone tool.
- Rank causes with transparent evidence and confidence scores, not black-box AI answers.
- Feed every experiment outcome back into the system so recommendations improve over time.
How it works
From input to outcome.
01
Performance and content data
02
Diagnostic and relevance engine
03
Content and recommendation generator
04
Automated experimentation
05
Compounding learning memory
One system that diagnoses, generates, tests and remembers, not four disconnected tools.
What it does
The capabilities.
Seasonal and commercial calendar
Holidays plus curated retail moments across 12 countries.
Vertical relevance engine
Matches every event to the specific verticals it affects.
Localised content generation
Headlines, ribbon copy and long-form copy in 9 languages.
Live AI search insights
Surfaces current market signals for any vertical and country.
AI-search content optimiser
Rewrites existing copy against an AI-search-ready checklist.
Search, filter and CSV export
Fast filtering with one-click export for planning cycles.
The result
What changed.
Localised at scale
Generates ready-to-use headlines and copy in 9 languages across 12 countries automatically.
Built-in relevance
Matches every calendar event to the specific verticals it actually affects, with no guesswork.
AI-search ready
Scores and rewrites existing copy against an AI-search-readiness checklist.
What could this become?
- Anomaly detection and diagnosis
- Ranked causal recommendations
- Automated experiment briefs
- A compounding learning memory
- A unified performance data layer
- Full article generation
Facing something like this?
Describe the process in a short note. Every system here started with exactly that.