The travel industry seems to be one of the most technologically advanced sectors. Prices are determined by algorithms. Search and booking processes are optimized down to seconds. Billions are invested annually in artificial intelligence, personalisation and conversion.
Most of this investment is well justified. Search, pricing and demand forecasting are indeed complex tasks, and the industry has made real progress in this area. But after booking, the situation changes radically.
Much of the work in the travel industry still relies on manual labour behind modern interfaces. This is where costs quietly accumulate, margins shrink, and scaling becomes difficult. Not because the industry lacks innovation, but because innovation is unevenly applied.
Where Costs Actually Accumulate
For airlines, OTAs, and TMCs, post-booking servicing is consistently one of the most expensive operational areas. Industry research from McKinsey and Skift shows that 30-40% of operational costs in agency and managed travel models come from what happens after the ticket is issued: exchanges, refunds, involuntary changes, schedule updates.
These cases represent a minority of total bookings, but they consume a disproportionate amount of effort. A typical manual exchange or refund takes 8-12 minutes of agent time. More complex cases easily exceed 30 minutes. Where automation exists, the same operation can be completed in seconds.
At scale, that difference turns into millions in recurring costs.
High-Frequency, High-Risk Manual Workflows
The most resource-intensive processes include ticket exchanges, refunds, waivers, involuntary changes, and multi-PNR itineraries. These are not exceptional cases; they are routine outcomes of an industry exposed to continuous schedule changes, weather disruptions, and operational volatility.
The financial risk is equally material. IATA consistently highlights Agency Debit Memos (ADMs) as a major source of revenue leakage and operational friction for airlines, with servicing and ticketing errors cited as one of the most common underlying causes rather than fraud.
Even a 1% error rate in manual servicing can generate substantial annual losses through penalties, rework, and customer compensation.
Why Manual Processes Persist
It is easy to blame slow implementation or resistance to change. In reality, manual work persists because of how travel infrastructure is structured – and how difficult it is to replace.
For small and mid-sized travel companies, building robust post-booking automation is complex, expensive, and risky. The simpler and more predictable solution is to absorb operational complexity with people. What starts as a few support agents gradually turns into teams of dozens or even hundreds as the business scales.
The core systems were developed decades ago. The logic of the EDIFACT era still underpins modern APIs. GDS, NDC and airline direct connections follow different service models. Fare rules are often published as long text documents rather than machine-executable logic. Service workflows vary depending on where and how a booking was created.
In such an environment, systems struggle to interpret intent and safely execute changes and people fill this gap.
Agents are not used because they are cheap. They are used because they can adapt when systems cannot. They act as a bridge between fragmented architectures.
The Hidden Cost of Human Dependency
The economic impact of manual operations extends far beyond wages.
Operating costs increase directly with the volume, limiting profitability. Response times suffer, leading to customer loss and brand dissatisfaction – a trend consistently reflected in customer reviews on major travel platforms.
It typically takes 3-6 months to train a booking agent to reach full productivity, while annual turnover in operations teams often reaches 30-50%, creating a continuous cycle of re-hiring and retraining.
Errors in complex reissues or refunds have direct financial consequences, including ADM penalties and reputational damage.
Why AI Hasn’t Solved This
Most modern AI-based tourism projects tend to remain superficial, with companies implementing AI through chatbots, self-service portals, and automated responses. These improve customer perception but do not change the economics of operations.
Language models can explain policies. They can direct customers. But without deterministic execution logic, transaction security, and auditability, they cannot reliably perform high-risk service workflows.
AI struggles in operations not because it is not powerful enough, but because the systems it connects to were never designed to run autonomously.
Breaking the Innovation Paradox
The paradox is simple: the more the industry invests in front-end innovation, the more money it continues to lose in legacy backend operations.
Solving this problem does not require a different interface or a different assistant. It requires a different foundation – normalised operational data, enforceable rules, transaction-safe coordination, and artificial intelligence systems built to work, not just advise.
Real innovation in travel will not be measured by how intelligent interfaces appear, but by how little human effort is required to keep core operations running.
In an industry with thin margins and high volatility, the most expensive choice is no longer technical risk. It is doing nothing.
The Costliest Decision Is Inaction
This gap is already costing money today. Airlines with complex networks, large OTAs handling millions of post-booking interactions, and corporate travel programs operating at scale are the most exposed. As travel volumes continue to grow while disruption becomes the norm rather than the exception, manual servicing costs increase faster than revenue.
Over the next 12–24 months, this imbalance becomes critical: higher volumes, tighter margins, rising labor costs, and growing customer intolerance for slow recovery create a point where incremental automation is no longer enough. Companies that continue to treat post-booking operations as a support function will see margins compress further. Those that treat it as core infrastructure will quietly pull ahead.
About the author
Nick Filatov, Founder & CEO, GDS42.AI is a tech entrepreneur and investor with over 20 years of experience building large-scale travel tech businesses. He founded and led one of the largest OTAs in Eastern Europe, scaling it to 9-digit GMV and millions of users. After stepping down, he shifted the focus to launching AI-first products in the travel and automation space – and to supporting a new generation of founders.
