Skip to content
  • There are no suggestions because the search field is empty.

When should you create a new project?

Create a new project when:

  • The country changes

  • The use case changes

  • The scoring logic changes

  • The brand changes

  • You need experimentation or training

  • You are developing a new GeoAI version

Important:
Each project supports only one scoring model. If scoring assumptions change, you should create a new project

Below are the most common and recommended use cases.

1. Country-Specific Projects (Most Common Use Case)

This is the most frequent setup for multi-country customers.

Create one project per country when:

  • Data availability differs by country

  • Scoring logic varies by market

  • Catchment assumptions differ

  • Local teams manage separate markets

Example:

  • Project: Germany Expansion

  • Project: Spain Expansion

  • Project: France Expansion

This keeps analysis structured and avoids mixing datasets or assumptions.

2. Different Use Cases (Operational Separation)

Create separate projects when workflows differ.

Examples:

  • One project for delivery area planning

  • One project for physical store development

Why this matters:

  • Delivery planning uses different catchments (e.g., drive time)

  • Store development may rely on footfall and retail POIs

  • Scoring logic differs significantly

3. Location Categories (Urban vs Rural)

Create separate projects when location types require different modeling logic.

Example:

  • Urban project → 5–10 minute walk catchments

  • Rural project → 10–20 minute drive catchments

Data points and scoring weights may differ dramatically.

This is particularly important for scoring because:

There is only one scoring model per project.

If urban and rural require different scoring logic, they must be separated into different projects.

4. Brand-Specific Scoring

If you operate multiple brands or concepts, create separate projects for each brand.

Example:

  • Project 1: Score Brand A

  • Project 2: Score Brand B

This ensures:

  • Clean scoring logic

  • No mixing of KPIs

  • Clear benchmarking

5. Template & Experimentation Projects

Many teams maintain:

  • One clean master template project

  • Separate duplicate projects for experimentation

Use cases:

  • Testing new scoring models

  • Training new colleagues

  • Trying new datasets

  • Scenario modeling

Best practice:
Maintain one stable “production” project and duplicate it when experimenting.

6. GeoAI Development Over Time

For GeoAI-based scoring, development often evolves year by year.

Example:

  • Project: GeoAI 2025 (currently in production)

  • Project: GeoAI 2026 (in development and iteration with new sales data)

This allows:

  • Stable operational use

  • Continuous model improvement

  • Parallel testing of new logic