Airlines and airports invest in their customer experience to differentiate and earn their passengers’ loyalty. When it comes to delivering this complex experience, they depend on a network of third parties: from baggage handling and wheelchair assistance, to cleaning, catering, and de-icing… every step on the journey requires every supplier to deliver the right service, at the right time, and to the right standard.
Optimising this complex ecosystem is one of the biggest levers airlines and airports have to protect margins and the passenger experience at once.
At a major U.S. airline, partners surfaced $6m in unrealized contractual value and overbilling as high as 35% from a single provider at a single location – not through bad faith, but because no one had a reliable way to see where contract, operations, and invoice had drifted apart.
Kubrick’s Verified Supplier Performance solution gives them a shared view of what’s actually happening on the ground, so all partners can act on it together.
Aligning suppliers behind the customer experience North Star
Partners want the same outcome: a passenger experience that meets the standard everyone agreed to. But no one has the evidence to have that conversation with confidence.
Business partner invoices vary. Some bundle every service line at a station into one bill; others send a separate invoice per service, per week. None reference the contract or SLA they relate to.
Elsewhere, an ops platform shows missed baggage last month. A report flags a cleaning contractor that fell just short of its turnaround standard twice this week. Neither signal reaches the invoice on the manager’s desk.
Reconciling contract, invoice, and operations means cross-referencing PDFs, spreadsheets, and reports, station by station, vendor by vendor. Faced with that, most airlines and their partners sign off, pay, and revisit it only if something looks obviously wrong.
A relationship without a shared source of truth
Running ground operations through third-party business partners costs on the order of a billion dollars a year for each airline. It’s one of the largest lines on the P&L, and one of the most important to get right for passengers. Yet most companies have limited shared visibility into how service delivery, contract terms, and billing line up.
Closing that gap means bringing together three things that usually sit apart:
- what the contract says should happen,
- what happened on the ground,
- and what was invoiced; typically, in different systems, formats, and teams.
Manual reconciliation is never going to scale!
Teams already check invoices against performance where they can, but doing it exhaustively takes longer than the invoice cycle allows.
There’s no shared identifier linking an invoice to its contract, so even matching the two requires interpretation: vendor name, dates, and service described, judged against a patchwork of MSAs, SOWs, and amendments, each worded differently. Invoice formats vary widely by provider too.
At scale, manual review falls behind the invoice cycle. Exceptions pile up, disputes drag on, and partners spend more time debating whose numbers are right than improving the service. Each team is doing its job well; what’s missing is a shared, trustworthy picture, not effort or intent. As companies add stations, service lines, and partners, that gap only widens.
The change: from self-reported to verified, together
Fixing this involves connecting contract, operations, and invoice into a single model, so what should have been billed and what was billed can be compared automatically, every cycle, at every station, and shared with the partner as a matter of course.
That’s the shift from self-reported to verified performance: a common record people trust, so conversations are grounded in evidence rather than opinion – recognizing strong-performing partners and giving underperforming relationships a clear and fair basis to improve.
Under the hood of the Verified Supplier Performance AI solution
At a high level, the solution comprises of:
- Ingestion: Contracts and invoices arrive as PDFs, and operational data as CSVs. Both land in a Databricks Unity Catalog Volume, then move through Databricks AutoLoader into a bronze layer.
- Storage: The dataschema is defined once, sitting behind a SQLAlchemy repository layer designed to run against Lakebase.
- Application: The solution is a Databricks App built in React.
- Processing: Bronze data is conformed into a silver layer using Databricks’ DQX data-quality checks. The reconciliation itself runs through a set of modules: multi-document billing netted across invoices, credit notes, and corrections over a period; a provenance/drill-down trace; a dispute state machine with audit events; PDF parsing combined with fuzzy line-matching and learned aliases; a versioned grading policy with on-demand alert evaluation; and temporal comparison.
The business outcomes
A shared, verified view of performance changes what every stakeholder can do:
- Resolve billing discrepancies faster, with both sides working from the same evidence.
- Automate invoice validation, so finance focuses on genuine exceptions, not every line.
- Recognize and invest in strong-performing partners, identified by verified data.
- Strengthen renewal conversations with a shared performance history, not anecdote.
- Protect the passenger experience, catching service gaps before they compound.
Beyond reconciliation: a foundation for stronger partnerships
Recovering value today is the starting point. The same connected model supports a longer-term shift: from explaining last month’s invoice to spotting supplier metrics beginning to slip before they affect a passenger.
It also extends beyond airport ground services, to any relationship built on third parties: aircraft maintenance, crew hotels, in-flight catering, facilities. A shared, evidenced view of performance is a capability that benefits the whole network — and every partnership in it.
