Scope
Contact centres generate a constant stream of interaction data — call recordings, email threads, chat transcripts, agent activity logs. At LOT Polish Airlines this volume exceeds 113,000 customer interactions per month, spread across voice, email, and messaging channels.
The challenge isn’t volume. It’s making that data mean something for the people running daily operations: Team Leaders deciding how to prioritise, Managers tracking service quality targets, HR identifying training gaps.
FlightDeck Insights is the analytics layer built to close that gap.
The Approach
Most contact centre reporting is agent-centric: how many calls did this person handle, what was their AHT, were they on time. Useful, but incomplete.
FlightDeck takes a conversation-first view. Each customer interaction is the unit of analysis — not the individual agent action. This shifts focus from activity metrics to outcomes that actually matter.
The system serves two distinct audiences: operational managers who need daily KPI snapshots, and analysts who need access to raw interaction history for deep-dive queries and audit trails.
The Architecture
Built entirely on Microsoft Fabric, with Data Factory handling automated ingestion from the Genesys Cloud API and Fabric Notebooks (PySpark) transforming raw interaction data into structured analytical tables inside the Lakehouse.
Power BI reports sit on top of the Fabric Lakehouse and refresh automatically. No manual extracts. No scheduled email distributions. The data is just there, current, every morning.
The reporting layer covers:
- Interaction archive — Complete copy of all Genesys interactions, structured and queryable for ad hoc analysis and compliance audits
- Service Level — Voice and email SL measured against operational thresholds, broken down by team, channel, and time of day
- First Contact Resolution — Conversations resolved without a repeat contact, tracked at agent and team level
- Interaction Volume — Monthly and daily trends, staffing alignment, peak identification
- Queue Health — Abandonment rates, wait time distribution, overflow patterns
What It Replaced
Before FlightDeck, performance reporting at the LOT Contact Centre was a patchwork: weekly Excel exports, manual pivot tables assembled by Team Leaders, KPI snapshots sent over email. The data arrived late, varied in methodology between teams, and couldn’t support real-time decision-making.
The shift to a live Fabric-based layer meant that when volume spikes or service level drops, it’s visible immediately — not on Friday in a summary report.