Case Study · AI Integration & Automation

Analysing 5 Years of ERC Grant Data with Claude AI

Five years of public ERC funding data turned into interactive dashboards in 15 minutes — using Claude AI and a Google Sheets MCP connection.

Tech stack
Claude DesktopGoogle Sheets MCPEC R&I Dashboard (Qlik Sense)CORDISChart.jsHTML
Final setup — what was built
Architecture
1

EC R&I Dashboard (Qlik Sense) — public database of all ERC-funded projects; filtered by grant type and call year, exported as CSV

2

Google Sheets MCP — connects Claude Desktop directly to the spreadsheet; Claude reads, cleans, and structures the raw export in context

3

Claude AI — performs cross-grant comparisons, institution ranking, domain breakdowns, and funding aggregations

4

Chart.js HTML dashboards — interactive, filterable, embeddable; one per grant type with country and institution drill-downs

5

CORDIS — cross-checks project abstracts, PI names, and institutional affiliations where R&I export fields were ambiguous

822
grants analysed (2021–2024)
~15 min
raw data to finished dashboards
4
interactive dashboards produced
01 · The real problem
The goal

Understand which institutions win ERC grants, which domains dominate, and how funding is distributed — without spending days in a BI tool.


The daily reality

The EC publishes granular data on every funded project but it lives in Qlik Sense — a platform most researchers and grant advisors have never used. Cleaning, analysing, and visualising the export has historically required data skills most people don’t have time to develop.

02 · Before vs After
Before
1
Navigate Qlik Sense to find the right dataset
2
Export and manually clean raw data
3
Build analysis formula by formula in a spreadsheet
4
Create static charts
5
Repeat for each grant type
After
1
Download filtered export from EC R&I dashboard
2
Claude reads, cleans, and analyses via Google Sheets MCP
3
Interactive Chart.js dashboard generated — all four grant types in one session
02b · The hidden misconception
Common assumption that's wrong

‘Detailed ERC grant data isn’t publicly available.’ It is. Every funded project is documented — institution, PI, country, funding, dates, abstract, CORDIS link — in a freely downloadable EC database. The bottleneck was never access. It was the capacity to analyse it at scale.

03 · Blockers and solutions
Blocker
Solution

Qlik Sense is not intuitive for first-time users — Non-obvious how to apply the right filters and export a clean dataset across multiple grant types.

**Apply all filters before exporting** — Qlik exports whatever is currently visible. Learn the filter panel first.

Raw export needs structuring before any insight emerges — Inconsistent institution naming and mixed grant types require cleaning before analysis is possible.

**Claude via Google Sheets MCP** cleans and structures the data in context — no manual formula work.

Four dashboards would normally take days to build — STG, COG, ADG, and SyG each have different structures and institution profiles — typically separate pipelines.

Claude handled all four in a **single session** with filterable tables, country breakdowns, and institution drill-downs.

04 · What Claude can now do

Four interactive dashboards covering STG, COG, ADG, and SyG for call years 2021–2024

Funding patterns identified: Germany and Max Planck lead STG/COG/ADG; CNRS leads Synergy; Life Sciences dominates across all types

Analysis time reduced from days to ~15 minutes

Reusable workflow — repeatable each trimester as the EC updates its R&I database

05 · What I'd do differently
Honest reflection

The data was always public. What changed is the capacity to work with it without a data team. Connecting Claude to a live spreadsheet via MCP removed the last barrier between public EC data and actionable grant intelligence — and the workflow is repeatable for any Horizon Europe programme.