19 May 2026 — Skymetrix today published the largest publicly available study of how airline invoices arrive at finance teams, and launched AI Invoice Automation, the new aviation-trained system built to close the structural gap the research exposes.
Our study finds that 40% of fuel and airport charge invoices still arrive on paper or PDF, requiring manual data entry. Combined with industry-wide error and verification benchmarks, that digitization gap points to an ‘invisible surcharge’ of $4.6 billion in undetected invoicing errors paid out by airlines each year — roughly an eighth of total industry net profit (IATA, June 2025).
AI Invoice Automation is our response: a touchless system that takes any invoice, in any format, from paper to paid automatically.
The issue: a digitization gap that has not closed
The study examines a single, format-level question: when an invoice enters Skymetrix’s processing gateway, does it arrive as a structured electronic message (EDI or XML) or as an unstructured document (paper, scan, or PDF that must be re-keyed)? Across 1,120,589 invoices observed at intake during 2024–2025, the answer is split:
- 40.3% arrive as unstructured documents — paper, scans, or PDFs.
- The industry is sharply polarised: some carriers operate above 90% electronic intake; others sit below 50%.
- Operational scale is not a reliable predictor of digitization — some of the highest-volume operators sit well above the industry average for paper.
The study reports format-level patterns only, in aggregate. No individual airline is identified, and no invoice content was examined.
“Airlines operate the most advanced machines on earth, yet 40% of the invoices those operations generate still arrive in formats that a computer cannot read.
“Once an invoice has to be re-keyed, error becomes statistically inevitable — and once it reaches the ledger, the cost of fixing it disappears into the noise. The ‘invisible surcharge’ is the financial consequence. AI Invoice Automation is how we close the gap.”
— Michael Scheidler, CEO, Skymetrix
The cost: an 'invisible surcharge' of $4.6 billion a year
Once an invoice arrives on paper, the only way to capture its detail is to re-key it by hand. The cost implications of that manual step are drawn from well-established external benchmarks and from Skymetrix’s broader experience over 25 years of working with airline finance teams:
- Manual data entry is error-prone. APQC’s Accounts Payable benchmarking puts the manual data-entry error rate at 3.6% — roughly one in every 28 fields entered from a paper invoice will be wrong.
- Invoices themselves contain errors at high rates. The Institute of Finance and Management reports that 39% of all invoices contain errors: wrong prices, quantities, surcharges or tax.
- Detailed verification is the exception, not the rule. Skymetrix consistently observes that AP teams verify only 10–20% of invoices line by line; the rest are approved on header total alone.
Skymetrix has consistently observed invoicing error rates equivalent to at least 1% of direct operating costs across 25 years of work with airlines. Applied conservatively to the industry’s $455 billion direct operating cost base (IATA, June 2025; EUROCONTROL Aviation Intelligence Portal), the resulting ‘invisible surcharge’ reaches $4.6 billion a year — roughly an eighth of total industry net profit.
Dive deeper into the numbers
Read the full paper detailing our research, analysis and findings. Based on 1.12 million invoice records observed at platform intake across 31 airlines.
The solution: AI Invoice Automation
AI Invoice Automation is Skymetrix’s answer to the gap. Unlike generic OCR tools that convert text and hand the output back to a person, the system runs four stages end-to-end:
- Read — Every invoice is ingested regardless of format: paper, scanned PDF, email attachment, EDI or XML.
- Interpret — The system recognises aviation-specific entities including tail numbers, flight numbers and charge types.
- Normalize — Every entity is matched against live operational data so the invoice is structured correctly before validation.
- Validate — Every field is checked against agreed contract rates, then either routed straight to payment or flagged as a genuine exception.
Clean invoices flow directly to the journal with zero human intervention. Exceptions — wrong prices, duplicate charges, contract drift — are surfaced before payment, not discovered months later in audit. Accuracy reaches levels manual checking cannot match, on every invoice, every time.
“Most invoice tools stop at OCR — they read the page and hand it back to a person. We built something completely different. Our system reads, interprets, normalizes and validates, all four stages, end-to-end, with no human in the loop unless there’s a genuine exception. That’s touchless invoicing, and it’s what makes the $4.6 billion problem solvable at scale.”
— Scott Henry, Director of Product Strategy Development, Skymetrix
Why it works: 25 years of aviation data, not a generic model
What separates AI Invoice Automation from horizontal AP automation tools is the data and domain expertise behind it.
Generic AI models have never seen an into-plane fuel ticket, don’t know that Jet A-1 pricing at Heathrow follows different conventions to a regional airport in Southeast Asia, and have no way to recognise that a 0.3% drift in a supplier’s unit rate over six months is a sign of contract erosion.
Skymetrix has been processing airline cost invoices for 25 years across 135+ airlines, including 7 of the 10 largest carriers in Europe. The platform has structured over $100 billion of real airline costs and the system validates against the data sources airlines actually use:
- contract rates
- fuel pricing indices (Platts, Argus, OPIS)
- uplift records
- flight activity
- tariff schedules, and
- FX feeds.
A generic OCR tool can tell you what a number says. Only an aviation-trained system can tell you whether that number is right.
“Anyone can build an AI model. The question is what you feed it. A general-purpose LLM doesn’t know that into-plane fees vary by handler, that tax treatment changes by jurisdiction, or that a supplier’s rate has drifted 0.3% over six months. We know all of that, because we’ve been processing it for 25 years across 135 airlines. The model is the engine; the data is the fuel.”
— Michael Charalambous, Chief Commercial Officer, Skymetrix
Methodology
The research observes the intake format of 1,120,589 fuel and airport charge invoices processed through Skymetrix’s gateway during 2024–2025.
The only data point recorded for the purposes of this research is whether each invoice arrived as a structured electronic message (EDI/XML) or as an unstructured document (paper, scan, or PDF).
Findings are reported in aggregate at industry level. No individual airline is identified, named or characterised, and invoice contents were not examined as part of the study.
All cost-impact figures are derived from third-party benchmarks (APQC, IFM, IATA, EUROCONTROL) and Skymetrix’s broader 25-year industry experience.
AI Invoice Automation
Read every line, of every invoice, for every flight. Download the one-page factsheet:
