Aviation

Contents

The whole map is here: the parts the book is split into, and inside them the chapters. Parts I have not started yet are listed as well, and the line under each chapter is a promise of what it will cover.

1. OriginsFrom a single seat on a 1925 mail plane to the 1978 deregulation: where today's data model came from.
2. 01From mail contracts to SABREaiMost of what looks strange in airline reservation systems today started with a single seat on a mail plane in 1925. This chapter is about why inventory and the passenger record were born apart, and why that split still costs us.
3. 02From SABRE to PSS: why one architecture lived 60 yearsWhy the 1964 data model still sits inside today's PSS, and why standardisation was missed on the first try.
4. 031978: when the profit guarantee ended, revenue management was bornWho decided fares, routes and inventory once the 55%-load, 12%-return guarantee was gone, and why revenue management became a survival skill.
5. Revenue managementThe discipline born when the profit guarantee ended: restricted discounts, controlled overbooking, leg control to O&D.
6. 04Yield management: the early strategy and its business logicHow restricted discounts, controlled overbooking and Littlewood's rule converged in DINAMO in 1985, and the formula for staying profitable while selling cheap seats.
7. 05Yield management: competitive strategy and the PEOPLExpress caseWhy a billion-dollar cost advantage was not enough; marginal traffic and inventory control in the losing side's own words.
8. 06Revenue management and strategic operations: PEOPLExpress and American AirlinesFrom leg-based to O&D control, the 30/70 math of hub-and-spoke, a cost culture down to the olive and the paint, and AAdvantage as a data tool, all in one formula.
9. 07PEOPLExpress and the industry: loyalty programmes and distribution systemsEvery mile redeemed displaces a paying passenger; the neutral shared system was tried five times and died five times; display order is a business rule.
10. 16Airline planning and revenue management: a strategic analysisFive links from schedule to profit: capacity planning, elasticity-based segmentation, O&D inventory control, CRM data carried into the booking moment, channel cost, and the competitor question no algorithm can answer.
11. 17Revenue management and competitive strategy in aviation: a business logic analysis through Sun Tzu's principlesTranslated into airline business rules, Sun Tzu's principles turn revenue management from a pricing tool into a competitive weapon that targets market share and margin together and demands clear direction, knowledge of both the rival and oneself, and decision authority with limits drawn in advance.
12. 41The airline industry and revenue management: an analysisWhy an extra passenger is nearly free when 80-90 percent of costs are fixed, why willingness to pay rather than cost sets the fare, why inventory maximizes revenue rather than profit, and the leg, segment, O&D service and market layers revenue management works on.
13. 42Airline revenue management alternatives and business logic analysisExplains the choice between leg-based and O&D inventory control, the data that feeds the RM engine, and why the two directions of forecast error do not cost the same.
14. 68Measuring airline revenue management performanceSplits RM performance indicators into pre-departure and post-departure measures, and explains why RASM and CASM must be read together, how spoilage and denied boarding costs are tracked, and how forecast error is interrogated with MAD, bias and WMAPE.
15. 69The revenue opportunity model (ROM) and measuring airline revenue management performanceHow ROM separates revenue management's contribution from the market by placing actual revenue between no-control and perfect-control scenarios, and how it splits the loss into spoilage and dilution.
16. DistributionMoving the reservation screen onto the agent's desk: who writes the standards, who carries the messages, who settles the money.
17. 08Airline reservation and global distribution systems (GDS): strategic evolution and business logicWhy MAARS Plus failed, what the four rules of 1984 banned, what MIDT and BIDT are for, and how four systems became three giants.
18. 09Aviation industry standards and governance: a strategic analysisIATA and A4A write the rules, SITA and ARINC carry the messages, OAG and ATPCO distribute schedules and fares, BSP and the clearing houses move the money; the full-content clause and the merchant-of-record question.
19. 10GDS and the airline distribution ecosystem: strategy and business logicThe GDS's customers are four clusters; how the 1994 e-ticket, the 2004 DOT sunset and the OTAs' demand for a thousand bookless results broke the mainframe; look-to-book from 10:1 to 10,000:1.
20. 11Airline reservation systems and digital distribution channels: a strategic analysisAssembly still beats at the heart of the system; the phased migration from TPF to open systems, and the shift of control from inventory to attention, from eAAsy Sabre through Priceline and Orbitz to Google Flights.
21. RetailingWho holds the value chain: agency economics, the commission cut, and NDC taking pricing power back.
22. 12The travel value chain and distribution channels: a strategic briefingRM decides, the host CRS executes; showing the same inventory in four storefronts is as critical as the RM math. 2.5 bookings per ticket, agency incentives above 50%, and the pricing power NDC wants back.
23. 13The travel distribution ecosystem and New Distribution Capability (NDC)The agency's five revenue streams, the commission cut of 1995, GDS surcharges since 2015; NDC moves pricing power back to the airline, ONE Order collapses three records into one, and the four certification levels.
24. 14NDC@Scale: transformation and business logic in airline distributionPricing power brought the compute load with it: 675 million searches a day, the GDS's existential choice, the gap the aggregators opened, and normalisation replacing transparency.
25. Fares and pricingFare products and their classification, rule engines, itinerary pricing, private fares, prorate agreements and ancillaries.
26. 18Airline pricing and yield management strategies: an analytical viewThe shift in 1978 from fixed fares to answering competitors, why marginal cost makes pricing reactive, the logic of systemwide versus market-specific changes, and the cascade that spreads through ATPCO, which carries 87% of fare filings.
27. 19Airline fare products and their business logicHow restricted, qualified and unqualified discounts differ, why revenue management breaks down once fences disappear, and how a fare is represented in the many-to-one mapping chain that narrows from fare basis code to booking class.
28. 20Classifying airline fare products: strategic analysis and business logicThe ATPCO categories (CAT 1, 15, 25, 35) that separate public, private and negotiated fares by who may sell and who may buy them, why a discounted corporate fare can cost more than a public one, and how the airline loses control of the final price when negotiated fares settle net through BSP.
29. 21Airline distribution channels and fare rules: a strategic analysisA ticket price sits where two layers meet: the full content agreement and its parity clause that constrain web fares, NDC making that agreement obsolete, ATPCO's 29 rule categories with the stopover threshold, and Fare by Rule cutting ADMs on corporate contracts.
30. 22Airline fare rules and journey types: a strategic analysisWhich pricing logic back-to-back tickets, hidden cities and point-of-commencement arbitrage exploit, how revenue integrity software catches them, and how journey types such as one-way, circle trip, open jaw and round the world are modelled in the system.
31. 23Itinerary pricing in aviation: business logic analysisHow a journey is split into fare components and recombined into priceable units, the open jaw distance rule, IATA TC areas, and picking the cheapest valid solution with taxes included.
32. 24Fare construction, segmentation and loyalty programs analysisFares built from gateway and add-on components, currencies reduced to a single unit via NUC, the number of price points bounded by inventory control, and the business logic behind stopover, open jaw and one-way redemption rules.
33. 25Special fares and price elasticity in aviationThe unrealized revenue a single price leaves on the table, private fares visible only to their target segment, the conditions behind bereavement and child fares, and price elasticity that falls as departure nears, along with how a competitor's fare moves your demand.
34. 26Fare management and planning strategies in aviationHow a single fare action is decided where the competitive landscape meets the market profile: responses that differ by competitor, price elasticity turning a discount into loss or gain, and the RASK goal split between volume and yield.
35. 27Reactive pricing process and strategic decision mechanismsThe five stages of responding to a competitor's fare move: detection down to rules and footnotes, impact analysis that weighs revenue dilution, when not matching is the right call, and the hard requirement to fit into the next fare distribution window.
36. 28Proactive pricing and fare rationalization: a strategic business logic analysisWill explain the rules behind proactive pricing decisions and why an RM system stops protecting higher classes when fares do not sit in a clean booking-class hierarchy.
37. 29Revenue sharing in aviation: multilateral and special prorate agreements (MPA and SPA)How a single ticket's revenue is split between two airlines: IATA's mileage- and cost-weighted default MPA, and the bilateral SPA that overrides it.
38. 30Airline ancillaries and their business logicHow services added back on top of a no-frills base fare split into EMDs and receipts, the move from flat fees to market-based pricing, carrying brands across channels with the ATPCO S-8 record, and total itinerary prices that change with loyalty status.
39. 31Airline revenue management and fare structuresWhich carrier and document fuel, channel and optional-service charges (YQ/YR, OB, OC) attach to, how unseen demand (spill) is estimated, and how recapture and upsell rates feed the inventory decision.
40. ForecastingBooking curves, untruncating censored demand, spill models and O&D-level forecasting.
41. 32The airline spill model and its business logicHow the demand a full flight cuts off is estimated, how six decisions from upgauging and cabin layout to corporate discounts and award tickets are priced from that one calculation, and why assuming an LFCF of 1.0 understates spill.
42. 33Expected spill and the Boeing spill modelEstimating the demand a closed flight never sees with the Boeing spill model: why demand variability doubles the loss at the same load factor and why the closing load factor pushes the tables up.
43. 34Aviation demand forecasting and spill models analysisExplains how the passengers a full flight turns away are estimated, why the Boeing model's logit approximation overstates spill at high load factors, and how the Gamma model closes that gap.
44. 35Expected spill and demand analysis in revenue managementShows how the passengers a full flight turns away are counted from the demand distribution, and why a single passenger's rejection probability always exceeds the flight's closing probability.
45. 36Calibrating input parameters for airline spill modelsHow the two inputs of a spill model, the demand coefficient of variation (CV) and the load factor of closed flights (LFCF), are calibrated from which data and with which filter, and why a small drift multiplies the estimate at high load factors.
46. 37Capacity management and spill (lost demand) analysis in aviationThe four assumptions behind estimating spill from the load factor of closed flights: Gamma versus Normal demand, a 96 rather than 100 percent LFCF, a CV that grows from 0.30 to 0.46 with scope, and an iterative calculation that converges on the observed load factor.
47. 38Nominal load factor and spill analysisIterating back to the true demand a full aircraft hides, the three fates of a spilled passenger, how ignoring recapture inflates nominal demand, and why First Class demand is modeled with a negative exponential rather than a bell curve.
48. 39High-variance demand and the two-stage Coxian distributionWhy first-class demand does not fit a negative exponential distribution, how the two-stage Coxian distribution opens up variance with a single transition probability, and how a 15-point difference in the load-factor-on-closed-flights (LFCF) assumption multiplies the spilled-passenger estimate.
49. 40Measuring spill with the two-stage Cox distributionRebuilding the demand the sales system never sees from the mean and deviation of history via moment matching: the a = 0.1 constant and its bound, how flight closing rate, expected spill and spill rate come from the same tail area, and how those numbers feed equipment and inventory decisions.
50. 43Revenue improvement and forecast accuracy: demand forecasting in two-dimensional timeWhy airline demand forecasting works in two-dimensional time (booking date and departure date), how unconstrained demand is rebuilt from observed sales and class open/close timestamps, how upsell and recapture enter the forecast, and how the forecast turns into leg/segment inventory decisions through total or remaining-demand formulations.
51. 44Booking profiles and demand forecastingHow bookings leading up to departure become a profile through reading days, and why that profile misleads the forecast unless closed-for-sale periods are separated out.
52. 45Clustering booking profiles and analysing cancellation ratesWhy a single flight's volatile booking profile is reduced to k-means standard profiles or hierarchical levels, why analysts trust the hierarchy, and why the cancellation rate profile, built as a share of bookings on hand, is the most stable input to overbooking.
53. 46Demand profiles and data cleansing in revenue managementCleansing booking counts before they reach the forecast: the net demand profile that absorbs cancellations to stay monotonic and the PNR data it costs, the cancellation rate profile too volatile to drive decisions, and untruncating a closed class's censored data back to unconstrained demand with open/close indicators and the standard booking profile.
54. 47Demand forecasting and unconstrained demand in revenue managementRecovering demand from closed booking periods with booking profiles and the EM algorithm, reducing it to net demand with cancellation and boarding rates, classifying holidays, and forecasting methods from time series to machine learning.
55. 48Airline demand forecasting and time series analysisHow demand is forecast from history before a flight has bookings, from moving averages to Holt-Winters and its α, β, γ parameters.
56. 49Forecasting models in revenue management and their business logicExplains when the Kalman filter, ARMA/ARIMA and regression work on airline booking data, and which business rules model selection turns into.
57. 50Airline revenue management: booking forecasting and demand analysisHow time-series and booking-profile forecasts are blended by days to departure, how demand hidden by closed classes is recovered with EM, and why a simple average often beats sophisticated models.
58. 51O&D demand forecasting: first- and second-generation approachesHow the move from first-generation forecast enrichment, which splits segment forecasts by historical ticket flows, to direct O&D forecasting fed by daily PNR data changed the data, the latency and the role of choice models.
59. 52Competitive airline shopping data analysisHow shopping data, which records what was shown to the traveller and at what price, feeds demand forecasting, dynamic pricing, schedule profitability, display ranking and NDC comparison through the rejected options MIDT cannot see.
60. 53Data-driven business logic and decision support systems in aviationHow shopping data feeds five decision systems: override commission targets revised by demand signals, filtering out robotic shops, buy-or-wait advice that flips by channel, O&D demand forecasting moving from QSI to CCM, and inventory control that dilutes revenue when it ignores upsell.
61. 54Airline revenue management and consumer choice modelingModeling demand as a choice rather than a number: designing bundles from revealed rather than stated preferences, the utility function and its β coefficients, MNL market share and the scale parameter, proportional re-attraction when a class closes, and why that assumption breaks when departure times are far apart.
62. 55Itinerary choice models and demand analysisExplains where demand flows by utility when a flight is removed or fills up, and why the IIA assumption of multinomial logit gets that shift wrong.
63. 56O&D forecasting and the must-forecast listWhy forecasting only the markets that carry most of the demand, and treating the rest as a pseudo-local residual on each leg, gives the network optimizer more consistent input than forecasting every O&D.
64. Inventory and availabilityOverbooking, nested classes, bid price controls, network optimisation, and whether the seat is really for sale.
65. 57Overbooking strategies and operational analysis in aviation and hospitalityExplains the balance between the cost of an empty seat and the cost of denying boarding, the quality-of-service constraints (premium cabins, EU 261) that pull that balance below the optimum, and why the same logic works differently in hotels.
66. 58Show-rate forecasting and overbooking strategiesHow the show rate behind every overbooking limit is computed: go-show and mis-connect corrections, the four-hour snapshot, PNR-based causal models, and a binomial search against a 20-per-10,000 denied-boarding cap.
67. 59Overbooking and show-up modeling strategies in aviationThe overbooking limit is set by the variance of the show-up rate, not its mean; explains why binomial and deterministic models are too aggressive, what the truncated normal distribution separates, and how the limit is chosen to maximize net revenue.
68. 60Overbooking strategies and revenue management in aviationExplains why the overbooking limit must be solved together with the fare mix, why the fall-off rate is a more robust input, and how the true cost of a voucher changes the calculation.
69. 61Discount allocation controls, Littlewood's rule and the Gamma demand modelExplains why a discount seat is sold by comparing today's fare with the expected revenue of a future full-fare passenger, and why protection levels should be computed with a Gamma rather than a normal demand distribution.
70. 62Gamma distribution and discount allocation: protection levels and revenue ratiosExplains why the protection level depends only on the demand distribution and the revenue ratio between classes, how the Gamma distribution is parameterised from historical data, and how rejected demand is won back through upsell and recapture.
71. 63Discount allocation and booking optimization: EMSR and integrated overbookingWill explain how the EMSR family decides how many seats to protect for high-fare passengers, why EMSRB earns more, and why merging it with overbooking promises 1-3 percent extra revenue.
72. 64Reservation inventory control and revenue managementHow non-nested, parallel nested and serial nested inventory controls do or do not stop a low-value class from blocking a high-value one.
73. 65Mixed and hybrid inventory control systemsMixed nesting that guarantees quota to low-yield traffic, hybrid nesting that isolates channels and forbids seat borrowing, and how threshold nesting derives availability from total seats sold.
74. 66Airline revenue management: inventory control and business logicNet and threshold nesting, SCIs and segment limits, POS controls: the rules by which a revenue management decision reaches the sales screen.
75. 67Shared cabin inventory and funnel flightsWhat counting one seat in two cabins and selling two flights under one number gains in load factor and visibility, and what it costs in inventory synchronization and during disruptions.
76. 70Critical state identification and O&D strategies in revenue managementHow KPI thresholds pick which flights deserve attention across 5,000 daily departures and 1.65 million inventory units, when analyst overrides add revenue and when they lose it, and why O&D control replaces leg-based control for connecting traffic.
77. 71Airline inventory control strategies and network effectsExplains why the same last seat sells for 225 dollars with no controls and 1,712 dollars with itinerary controls, and how a promotion on one route ripples through the rest of the network.
78. 72Airline revenue management: an analysis of virtual nestingHow thousands of itinerary classes are clustered into a few nested virtual buckets by the passenger's net network value (CER), and how inventory is controlled through those buckets.
79. 73Virtual nesting and dual indexing: who gets the seat in airline inventoryHow net value (CER), found by subtracting a leg-specific displacement cost from the fare, spreads seats across four virtual buckets regardless of booking class, why the same K class can be open for one O&D and closed for another, and why dual indexing, which values a connecting passenger separately on each leg, is enough in a network where 70 percent of traffic connects and the average trip uses 1.7 legs.
80. 74Dynamic virtual nesting and revenue management analysisExplains why the bucket an itinerary falls into is decided at request time, and why bucket boundaries should be drawn by dynamic programming that minimizes displacement cost rather than by equal traffic.
81. 75Airline revenue management and O&D optimization: virtual nesting and CERHow hundreds of O&D-class combinations are ranked by CER rather than fare and partitioned by dynamic programming into the eight virtual buckets a reservation system allows, and why the mapping must be fed with flown ticketed data.
82. 76Continuous nesting and bid price control systemsExplains how continuous nesting prices every seat by its opportunity cost instead of allocating it to buckets: the net contribution rule, bid prices that move with each sale and cancellation, linear gradients versus bid price curves, and how market variables like POS enter the decision.
83. 77Airline revenue management: network optimization modelsHow the model that maximizes expected network revenue under capacity constraints is built from Rsc, Dsc, Cj, Isc and Xsc, how bid prices emerge as the dual values of leg capacity, why deterministic demand produces aggressive controls, and how Lagrangian relaxation shrinks the computation on large networks.
84. 78Network optimization and leg decomposition in revenue managementThe assumptions a network optimization engine needs to get right: fast table lookups for Gamma demand, Lagrangian improvement and stopping criteria, nested inventory and bid price-based proration.
85. 79Airline revenue management and network optimization strategiesHow inventory is managed with bid prices under O&D control, why network optimization must cover a full day's schedule and rerun 5 to 50 times a day, and why dynamic programming gives way to hybrid models on large networks.
86. 80O&D revenue management and seat availability calculationExplains how a booking request passes a physical check and then a financial check against the total bid price, how the same product opens differently by point of sale, and how this logic is attached to a legacy CRS.
87. 81Fare qualification rules in passenger valuationWhich rules filter the qualified fare behind net contribution in O&D inventory, why NDC forces the market value table to give way to real-time pricing, and how the point of commencement values the return segment.
88. 82Airline revenue management and inventory control systems: a post-process nesting analysisExplains how network optimization output is translated into nested limits on a legacy host CRS, how satellite processors work around TPF/ALCS constraints, and how branded fare families replaced cryptic fare codes.
89. 84Airline inventory management and GDS integration systemsFour integration levels from AVS/AVN teletype to DCS/DCA: what the agent sees, when confirmation arrives, and why O&D control is only possible at the top level.
90. 85Airline inventory control and O&D managementWe examine the three mechanisms that make O&D control hold at the point of sale (market restricted flights, married segments and journey data) and the trade-off between GDS polling fees and control discipline.
91. 86Airline reservation and inventory management: strategic business logic analysisShows how inventory answers a seat request by looking at who asks, from where and as part of which journey, through Married to Journey, the POS hierarchy, POC and bid-price control of off-tariff fares.
92. Offers and merchandisingSegmentation, recommendation engines, bundling, dynamic pricing and the seat itself becoming a product.
93. 83Branded fare families and connectivity architectureExplains how a branded fare bundle is built, why price consistency between brands rests on an inequality, and why 30 booking classes do not fit into the GDS's 26-letter alphabet.
94. OperationsShopping traffic, screen display order, schedule planning and revenue management that survives disruption.
95. 15Airline marketing planning: process and business logicA planning loop that starts with the fleet decision five years out and runs to close-in re-fleeting three months before departure, the feedback between functions, and the blind spots of MIDT, IATA DDS and DB1B.
96. AI and what comes nextnot started yetAutomation, interpretability, digital identity, and the e-commerce giants moving into travel.