Level Up UK — Geo-Targeting Intelligence Tool

Crossreferences your community database with free UK open data to identify where paid ads will have the highest impact.

---

What it does

1. Pulls postcodes from your SQL database
2. Maps each postcode to an LSOA and parliamentary constituency (via ONSPD)
3. Enriches with:
   • ONS Census 2021: % women, % age 18–45, % degree-educated
   • 2024 General Election: progressive vote share (Labour + Green + LibDem + SNP + Plaid)
4. Scores each area:
   • Density score — how many Level Up members already live there
   • Opportunity score — demographic/political signal, discounted for areas already well-covered
5. Outputs:
   • scored_postcodes.csv — upload to Meta Ads or Google Ads
   • targeting_map.html — interactive map to explore visually

---

Setup

Requirements

pip install pandas numpy folium requests psycopg2-binary
# or for MySQL: pip install pymysql

Step 1 — Configure your database

Edit geo_targeting.py, find the DB_CONFIG block at the top:

DB_CONFIG = {
    "engine": "postgres",        # or "mysql" or "sqlite"
    "host": "your-db-host",
    "port": 5432,
    "database": "your-db-name",
    "user": "your-username",
    "password": "your-password",
}

POSTCODE_QUERY = """
    SELECT postcode FROM community_members
    WHERE postcode IS NOT NULL
"""

Also update POSTCODE_QUERY to match your actual table and column names.

Step 2 — Download the data

python download_data.py

This will auto-download Census 2021 and electoral data.

For ONSPD (the postcode directory), you need to download it manually because ONS changes the URL quarterly:

1. Go to: https://geoportal.statistics.gov.uk/
2. Search: ONS Postcode Directory
3. Click the most recent version → CSV download (~800MB zip)
4. Unzip and rename the main CSV file to data/ONSPD_raw.csv
5. Re-run python download_data.py to slim it down

Step 3 — Run

python geo_targeting.py

Outputs:

• scored_postcodes.csv
• targeting_map.html

---

Data Sources

Dataset	Source	License
ONS Postcode Directory (ONSPD)	geoportal.statistics.gov.uk	Open Government Licence
Census 2021 (TS008, TS007, TS067)	NOMIS / ONS	Open Government Licence
2024 GE Results	House of Commons Library	Open Parliament Licence

All three are free, openly licensed, and do not require registration.

---

Scoring Formula

Density Score (0–100)

Members per 1,000 residents in that LSOA, normalised.

Opportunity Score (0–100)

Weighted demographic index:

• % women aged 18–45 → 40% weight
• % with degree-level education → 15% weight
• % female population → 15% weight
• Progressive vote share (Lab+Grn+LD+SNP+PC) → 30% weight

Then discounted (up to 50%) for areas already well-covered by existing members.

High tier = strong demographics, not yet well-reached → prioritise for ads Low tier = already well-covered OR weak signal → deprioritise

---

Using the CSV with Meta Ads / Google Ads

Meta Ads (Facebook/Instagram)

1. In Ads Manager → Audiences → Create Audience → Custom Audience
2. Or use Location Targeting: paste the postcodes from the "High" tier rows
3. You can also upload the full CSV as a customer list to find lookalike audiences

Google Ads

1. Campaigns → Locations → Bulk add locations
2. Paste postcode list from the "High" tier

---

Extending the tool

Ideas for future iterations:

• Add Indices of Multiple Deprivation to weight by class/income
• Pull Twitter/X follower geodata (if available) to validate the model
• Add ONS Subnational Population Projections for growth areas
• Build a Flask/FastAPI endpoint to refresh scores automatically on a schedule