Industries / Transport and vehicles / Automotive / Car rental
Reservations and pricing
This is how a renter finds and books a car, and how the company sets the price for each car type, place and date.
5
Milestones achieved, from 1972 to 2013.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1972
A computer takes the bookings
Avis introduced Wizard in 1972, which it calls "the first real-time information management and reservation system." By Avis's account, Wizard reached Europe in 1985, linking more than 2,000 Avis locations and able to work with "all major reservation systems," and ran in 53 countries by 1996. Avis's Australian site still asks customers for their "Wizard number." Hertz's 2025 annual report still lists global distribution systems, the networks that let travel agents connect to a renter's reservations, among its sales channels.
Credit
Avis
1975
No car for a dollar
In a cease-and-desist order of July 31, 1973, the US Federal Trade Commission found that the trade name "Dollar-A-Day" misrepresented the price of Resort Car Rental System's cars, since mileage, insurance and other mandatory charges meant no car could be had for a dollar a day. The company had to drop the name. On April 14, 1975 the Ninth Circuit court of appeals upheld the order, holding that the law "is violated if it induces the first contact through deception, even if the buyer later becomes fully informed before entering the contract." Fifty years on, the FTC's 2025 rule requiring all-in prices covers only live-event tickets and short-term lodging, though its record cites the Dollar-A-Day case and actions against rental car fees.
LessonThe advertised price must be one a renter can actually get.
Credit
Federal Trade Commission; US Court of Appeals for the Ninth Circuit
1994
Models rescue National
In 1993 National Car Rental faced being wound up, and its owner, General Motors, took a $744 million charge on it. National built revenue management: a set of computer models to manage how many cars it had, what it charged and which reservations it accepted. A first version in July 1993 returned National to profit, and a fuller system from July 1994 added $56 million of revenue in its first year, by the account of M. K. Geraghty and Ernest Johnson, who worked at National, in the journal Interfaces. General Motors sold National in April 1995 for an estimated $1.2 billion.
Credit
National Car Rental; M. K. Geraghty; Ernest Johnson
2010
Brokers resell the world's rentals
On May 18, 2010, Priceline bought TravelJigsaw, a Manchester car hire broker founded in 2004 that booked cars at more than 4,000 locations in 80 countries, with customer support in 20 languages. A broker sells other companies' rental cars and passes the booking on. Priceline, since renamed Booking Holdings, reserved 88 million rental car days in 2025, up 5.8% on 2024. Hertz says such third parties, from online travel agencies to airlines and credit card companies, bring "a significant amount" of its bookings.
Credit
TravelJigsaw (managing director Gregory Wills); Priceline
2013
Later pages do not excuse
Abel Rent a Car in Brisbane advertised cars at $35 a day, then added a $5 daily "vehicle registration recovery fee" and a 7.5% "administration fee," so the true cost was $43.54 a day. After action by the Australian Competition and Consumer Commission (ACCC), the Federal Court ordered the operator, Nonchalant Pty Ltd, to pay $30,000 in penalties, as the ACCC announced on June 19, 2013. Justice Gordon held that showing the extra fees on later pages of the booking was not enough to undo a misleading price on the home page.
Credit
ACCC; Federal Court of Australia
Watch for
Renting a car by conversation
On May 25, 2026, Booking.com announced an app inside ChatGPT where travelers describe the car they need in plain words and get offers, along with an AI helper that answers questions about payment, pickup and insurance at any hour.
Achieved whenAn outside test shows that the offers and answers match the rental terms and the price renters pay at the counter.
Claimed by
Booking.com, May 2026
Open
01
Check the AI rental helpers
A test set of about 200 questions renters actually ask, about deposits, damage excess, fuel policy, young driver fees, crossing borders and one-way returns, each with the right answer and the clause it comes from in the company's own rental terms. Score the AI assistants that booking sites and rental companies now offer, and publish every wrong answer.
The briefCredit · unclaimed02
An open rental pricing simulator
A free simulator of one rental company's bookings and fleet, with cars in three or four types, reservations made days or weeks ahead, rentals of different lengths, one-way returns, upgrades and cars joining and leaving the fleet. Anyone could test a pricing or fleet-planning method on it and compare results with simple published baselines.
The briefCredit · unclaimed
Open 01 · the brief
Check the AI rental helpers
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A test set of about 200 questions renters actually ask, about deposits, damage excess, fuel policy, young driver fees, crossing borders and one-way returns, each with the right answer and the clause it comes from in the company's own rental terms. Score the AI assistants that booking sites and rental companies now offer, and publish every wrong answer.
Why it is still open
Booking.com put an AI rental helper and a ChatGPT app in front of renters in May 2026, and no outside test of their answers turned up in this research. A September 2026 column in Auto Rental News, by the president of the rate-tracking firm RateHighway, mentions an industry white paper on how closely AI-quoted rates match direct rates, but names neither its author nor its results, and it concerns prices, not answers about rental terms. Price display is still a live problem: in January 2026 the ACCC said EconomyBookings, a rental broker, had paid $39,600 over two notices alleging it did not show the total price as a single figure and represented that a pre-selected optional extra was included in the price when it was not.
Built with what exists
The rental terms that each company publishes for each country; the assistants under test, used only up to the point before payment; a spreadsheet of questions, answers and clause references; two people checking every scored answer.
First step
Download the rental terms of three companies at one busy holiday airport and write 30 questions with their answers.
Done looks like
200 questions covering at least three companies at ten airports in four countries, at least three assistants scored, and a published error rate for each with the wrong answers listed.
Open 02 · the brief
An open rental pricing simulator
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A free simulator of one rental company's bookings and fleet, with cars in three or four types, reservations made days or weeks ahead, rentals of different lengths, one-way returns, upgrades and cars joining and leaving the fleet. Anyone could test a pricing or fleet-planning method on it and compare results with simple published baselines.
Why it is still open
The open fleet simulators that researchers share are built for on-demand rides and ride-hailing, such as FleetPy and OpenHail, the latter released in September 2026. The textbook car rental problem, Jack's Car Rental from Sutton and Barto's reinforcement learning book, is available as an open Gym environment, but it models two locations and cars moved between them overnight, at a fixed $10 a rental, with no prices or advance bookings. Europcar's Opticar combines forecasts, simulation and optimization for its revenue managers in nine countries, but it is Europcar's own system, built with ACT Operations Research. No open simulator built for car rental, with its bookings made days or weeks ahead, turned up in this research.
Built with what exists
Python and the Gymnasium interface used for reinforcement learning environments; Google's OR-Tools for baseline planning methods; seasonal patterns from the US Bureau of Labor Statistics' car rental price index.
First step
Write a one-location version with three car types and a 90-day booking window, and test a fixed-price baseline on it.
Done looks like
The simulator released under an open license with five test scenarios and three baseline methods, and results reproduced by at least one team outside the authors'.