A History of Failure, Ep4

CONFIRM RS

Every villain has an origin story. This is mine.

I tell people that the first project I ever was a part of failed so hard that it has its own Wikipedia page and has been taught as a case study at Stanford. It is a tale of mismanagement writ large on the multinational stage. It includes lying, subterfuge, and other ethics violations. It’s one thing to fall behind and own it. It is another to tell everyone on the Titanic that the ship is not sinking. So, forgive me if I am more snarky than usual. After all, I was there, watching it all fall apart in front of me. But like Celine Dion, my heart did go on. So, grab some coffee kids, cuddle up, and listen to Uncle Robin’s story of CONFIRM RS.

In the Beginning, There Was the Ship

To do so, we need to go back in time to 1987, back when God and I and dirt were young. I was working at the Budget Rent a Car worldwide reservation center in Carrollton, Texas. We had a reservation system, known as NORTH, which worked for us, although by the time I started, it was getting long in the tooth. And by that, I mean it was at least 10 years past its “holy crap, we need something new” date. Directly across the parking lot from our humble work abode stood the Hilton Hotels reservation center, which also used the same reservation system tailored for them. Hilton was having pretty much the same issues. I know, I worked part-time there as well.

If you zoom out a few miles you would also find the reservation center for American Airlines between Dallas and Ft. Worth, south of the airport. American used a system called SABRE, as did every travel agent worth their salt worldwide (Travel agents. How quaint.) Zoom out further and you have Marriott Hotels.

One day in 1987, someone at American said, “Gee, wouldn’t it be a great idea if you go to one place and book a plane, car, and hotel reservation in one sitting?” And really, it was a great idea, because the internet had yet to be invented and things like Priceline and Expedia had not been thought of yet. And so, the slog began. American first approached Marriot with the idea. AMRIS (the IT arm of American Airlines) would take on the build and management of the system, called CONFIRM, and the users would pay for the service. Now, Marriott’s system, known as MARSHA was considered state of the art at the time, and the only reason why they considered it was if, as the company’s vice President stated, “the joint venture could produce a reservation system that was functionally richer than what they had and would cost less to operate.” That is corporate speak for “Give us a couple of bells and whistles and a reason to fire most of our IT department.”

Once Marriott was on, it was a few weeks before Hilton and Budget joined in, creating a group called Intrico. That office was across the street from both Budget and Hilton reservation centers. The brain drain was on. Technical experts were needed to help develop the system, and of course, some of the better floor managers were siphoned off by the lure of better pay and the promise of a great beyond. May 1988, saw AMRIS issue a press release telling the world that the design process had started. By September 1988, the agreement was finally hammered out. It was quite simple. AMRIS would:

• Design, develop, operate, and maintain a “State of the art” reservation system to be marketed worldwide.
• Would design and develop the interfaces so users could get everything done in one space.
• Market the system to others for profit
• Convert each of the partners' reservation systems to CONFIRM

The four partners paid $55 million and agreed on a deadline of June 1992. AMRIS assured everyone that everything was great and that the job would be done so that the operational costs would be around a dollar per reservation. The project would be in two stages: the design phase (7 months), and the development phase (45 months).

All Aboard the Titanic

And that is the moment everything started going wrong. You are designing a system that not only needs to talk to three separate multinational entities but eventually works with those entities’ mainframe systems (Mainframes were all the rage in the 1980s). And you think you can design it in seven months. I can see managers at AMRIS saying “A reservation is a reservation is a reservation.” Not so fast, Gertrude Stein. Even being a low underling at Budget Rent a Car, I knew that there were all different types of car reservations. I knew that it was the same with hotels as well. Your system has to take all of the variables into account. And we are not talking about one or two variables. Try to rent a car in London, England that a handicapable person can drive. Oh, and they need a car seat. Yes, it happens and more than one thinks. And we’re just talking cars. There are also two other multinational hotel companies and an airline involved. Seven months was a wish. How should have this been handled? This was a case where you present what you have, it gets marked up by your Subject Matter Experts, and you go back and update and present again until everyone agrees. Daily. This is building the framework the development hinges on. But here is the thing: The design team only met with the SMEs once a month.

In December, three months later, AMRIS presented a “base design” of the system to Marriott. Marriott said, “Are you kidding me with this?” The document described the functionality but provided no detail. It was like saying that you were going to Paris, but not how you would get there, or where you would be staying, or what you would be doing. This is now one year since the initial proposal. So, AMRIS went back and worked on it some more.

Is That an Iceberg?

In March 1989, AMRIS declared the functional and technical specifications were complete. They then started to circulate a preliminary plan for development, which was met with a unified “NO”. The next six months were left to hammer out all the requirements. If you do the math, the schedule was already off by 7 months. What is a company to do?

Why, increase the cost to $72.6 million and the cost of each reservation to $1.30. Now this is where things get interesting. You see, in the contract, this was the point of no return. According to the contract, the three partners could say adios when the development plan was presented.

Remember I mentioned lying and ethics? Yeah, in August 1989, AMRIS met with the three partners to review the financials of the project. The financials presented by AMRIS were a lie. Marriott was to find out two years later that AMRIS overstated the number of reservations presented and the actual operating cost was more like $2.00 per reservation. Cost per reservation is crucial in the hospitality industry. The real amount would have probably sent all three heading to the exits. Based upon the false information the trio opted to stay. The development plan was accepted and the deadline was pushed to July of 1992. AMRIS once again assured everyone that the Titanic was unsinkable and that we had only a minor scrape with the Iceberg.

The contract outlined four major development phases: Business Area Analysis (BAA), Business System Development (BSD), Coding, and finally Testing (Enter your humble narrator). Testing in this case was a little bit of a misnomer. Not only would the testing team be responsible for making sure every little oddity was covered, but we were also responsible for the conversion of a year’s worth of reservations, a number well north of a couple of million. And all of this had to happen before we went live in June 1992. To be honest, I was chosen to head up the testing team because no one else wanted it. You see, most of us on the reservation floor were hearing from our friends who went across the street initially. There were horror stories of delays, inept management and money being spent on things like Aeron chairs and a sculptural glass receptionist desk that looked like belonged in an art museum. It sounded like a disaster and we were still three years out. But you only die once. So, the kid with OCD and four perfectionists started to casually assemble once a week to figure out what we would need to do.

Why Are We Filling Up With Water?

In the meantime, January 1990 came and with it the news that AMRIS had missed the contractual deadline of the terminal screen design. In February 1990, AMRIS missed the completion milestone of the BAA phase. It was at this point they admitted, “Yeah, we’re 13 weeks behind, but we’re still good.” In March, the company began a six-week “replanning” effort, which is corporate speak for “Oh, Shit, time to circle the wagons.” There was a reason.

In the summer of 1990, both Marriott and Budget expressed concerns that the project was behind schedule and that the management was ineffective. According to later court records, the project office knew this. Our friends across the street knew this. They were wondering if they could come back to their old jobs. Also according to later court records, AMRIS management told the team to change the revised dates to reflect the original project calendar. Marriott asked for deliverables and was ignored. In October 1990, the partners were told the project was one year behind schedule.

But we’ll still make the launch date. Yeah. Right. The ship was not sinking. Everyone stay calm.

Meanwhile, the money kept coming in being billed at around 1 million a month from each of the partners. My team was told to keep meeting because our information would still be needed in one way or another. If that doesn’t give you a sinking feeling, you’re not watching.

Why Are People Heading To the Lifeboats?

February, 1991. AMRIS announced another “Re-Plan”. According to the new plan, only Hilton would use the system by June 1992. And this is the point, dear friends, when the knives came out. Marriott was told they would not receive their features until March 1993. Marriott had also been listening to the internal grapevine quite closely and blew the lid off, telling the world that AMRIS knew damned well the new schedule could not be met and had forced employees to change their timetables about the project; those who refused were either fired, resigned or sent to other projects. Marriott also started prepping their lawyers. Budget and Hilton followed suit. Why?

The price of the project had been increased to $92 million.

The summer of 1991 was a bloodbath. AMRIS employees knew they were fucked. They saw that even if they were working 9-hour days seven days a week, this was not going to be finished in time. People started quitting. A consultant was brought in. His report telling AMRIS that the ship was sinking was “buried” by senior management and he was dismissed. In October, the president of AMRIS decided to spend more time with his family and left the company.

Yet, Hope, or the sunk cost fallacy (take your pick), still sprang eternal. Marriott and Hilton decided to give the group one more chance, even though their own evaluations concluded it wasn’t going to happen.

Cue Celine Dion…

And so, in February 1992, my team was asked to take a walk across the street concerning testing. We were hopeful; I mean we had all of our test scenarios ready to go. We arrived early in the morning. We were escorted to a room that had a full-spread breakfast ready for us. We had a seat. There was a dog and pony show about the system that took a couple of hours. And then…

We were handed a stack of paper showing what they wanted to test and what the result would be. No system to sit down in front of. Nothing real, only really bad copies. Our test scenarios were not even looked at. We were basically told that they had done the testing for us. Lunch was brought in. When asked about existing reservations, we were told that they would be taken care of. When asked about what would happen to reservations that may be rejected and how that would be handled, the answer was not to worry.

The ship was not sinking. It was merely tilting at 45 degrees.

And I had to tell the best damn leader I had ever worked for what just happened. The response was “I know” and a deep sigh. At least I could keep my job as a floor supervisor.

Hilton, though, got a shot in April to beta test. It did not go well. That’s an understatement. You see, in the reservation center world, the average call time is important. For Budget, if the average call time went up by 1 second, that was an increase in yearly costs of close to a million dollars. The average call time for a reservation is about a minute and a half. This was taking multiple minutes to do one simple scenario. My sources across the parking lot say that the screaming from the executives on the top floor could be heard all over, due to the open architecture of the building.

If you weren’t in a lifeboat by that time, consider yourself Leonardo DiCaprio in the water. Cue the ship’s band playing “Nearer My God to Thee”. The project was going down, hard.

On April 29, 1992 the AMRIS chairman admitted as much, and told the partners that their best estimate was 15 to 18 more months. There was also the promise to repay 100% of the investment for anyone who wanted to pull out. The day before, AMRIS fired 8 top executives and replaced another 15 employees. A few days later, the Vice Chair admitted that the system interfaces and databases were inadequate to provide the necessary performance and reliability, something that should have been noted a few years before.

In July 1992 and $125 million dollars (280 million and change in today’s dollars) Intrico broke up and the lawyers took over. On January 1994, AMRIS reached settlements with the three partners

CODA

Budget still needed a new reservation system. After talks, it was decided to pair up with Avis, as their system was top of the line. I was asked to get the band back together. We became the User Group and were an intricate part of just about every step. It took about 18 months to tailor their system to our needs, but we ultimately succeeded. Of that, I am more than proud.

What did I learn? Listen to the people on the ground. While the executives may say one thing, the folks who are doing the work are going to tell you the truth. After all, we were hearing the stories well before anyone else. Second, don’t lie. Take ownership. If you lie once, then you’re going to lie again to cover up the first one. This was less simple miscommunication and more of a grand deception. And it taught me to always question. Always. That’s how this villain has rolled since. Question everything.