Automated Identification of High-Potential Solar Rooftops

Solar sales is expensive — and most lead lists are guesswork

Solar installation companies burn time and budget chasing rooftops that don't pan out. Reviewing every prospect through Google Maps, gut instinct, and site visits simply doesn’t scale, and it leads to inconsistent decisions across the team.

A modular data pipeline that scores rooftops by
solar potential

Disparate geo and solar datasets are merged, enriched, and ranked into a clean, actionable lead list — including annual energy yield, obstruction detection, PV status, and export-ready output for CRM and sales.

Dortmund: A Full End-to-End Pipeline Run

A complete pipeline run covering the city of Dortmund — from raw cadastral data through solar and AI enrichment to a ready-to-use prioritized lead list.

What this means for
solar sales teams

The system improves pre-qualification and reduces wasted sales effort. It's not a replacement for final technical planning, but it gives teams a faster, more consistent starting point for deciding where to focus.

Actual impact will vary by region, data quality, and sales process.

A filterable web report for
solar lead discovery

Filter and rank rooftops based on data that has already been fully analyzed. The report supports the full outreach workflow: identifying high-potential prospects, planning campaigns, and handing off clean lead lists to the field.

• ✓ Minimum annual yield (kWh/yr) and panel count thresholds
• ✓ Max-obstructions filter + toggle for "no existing PV"
• ✓ Filter by orientation, tilt, roof condition, and material
• ✓ Ranked output — not a raw, unsorted building list
• ✓ CSV export ready for CRM workflow and sales team handoff

Interactive potential map
for territory planning and lead clusters

The map view adds a spatial layer to the web report, making potential clusters, available roofs without PV, and priority areas visible at a glance.

• → Color-coded by potential tier (e.g. kWh/yr)
• → "No existing PV" filter to focus on new customer opportunities
• → Spatial clusters easy to spot
• → Supports campaign planning and field team routing
• → Territory-level view (map) and individual lead view (report) — in one place

Technical decision data
for each rooftop

Beyond lead screening, the system provides key technical metrics for each rooftop to support a reliable first-pass assessment.

• → Address / property location
• → Roof area and usable roof segments
• → Orientation and tilt
• → Solar irradiation / solar potential
• → Estimated annual yield (kWh/year)
• → PV status, obstacles, material, and condition

Data Pipeline & Enrichment
(modular, testable, city-agnostic)

A clean, modular ETL stack with clearly separated stages. Each phase can be tested independently, re-run as needed, and deployed to new geographies without rework. Multiple data sources converge into a consistent output schema.

The System Architecture
at a Glance

Geodata, the Solar API, and Vision AI feed into a single modular pipeline. The output isn't just an analytics model — it's a deployed, usable system: reports, an interactive map, and data exports ready for sales operations.

Transparent.
Explainable. Built for sales teams.

This isn't a black-box ML score. Every component is visible, weighted, and easy to explain — which matters when sales teams and planners need to trust the score and act on it.

What this project demonstrates

An end-to-end data product: from problem definition and architecture through implementation, deployment, reporting, and operational use.