How We Did It: The Case of the Misconnecting Passengers

How We Did The Investigations

The Case of the Misconnecting Passengers

© Teradata 2011

We’re Getting A Lot of Questions …

Hi Everybody,

We wanted to answer your questions about how we did our brainstorming to help out AirLondon in the Case of the Mis-Connecting Passengers.

This write-up will give you an idea of our clients’ architecture and some details from our investigation.

Take a look, and if you still have questions, shoot them to us! We’re all on Facebook.

Yours truly, Neuman Hitchcock Chi Tylana Mathieu Ames

BSI Teradata Presents … The Case of the Misconnecting Passengers

Case in progressYou can help!

© Teradata 2011

• AirLondon has been in business for 3 years now• You work in the Operations group, and are responsible

for handling mis-connecting passengers. • BSI has been hired to help build a new Rebooking

Rule Engine, using Teradata

• Word just came in that there is a new Misconnect Situation at our hub in Frankfurt. 4 people are trying to get to London from various parts of the world. There are 2 available seats right now on the next and last flight of the day on AirLondon.

• You must pick the 2 people who get to go to London, and the 2 who stay overnight in Frankfurt

The Case

© Teradata 2011

The Problem At Frankfurt

We didn’t have to do any work here – you can do a Google search to find the Airport Trackers for most hubs, including this one for Frankfurt. The planes report their locations every 15 seconds so at all points in time AirLondon can load this information into their Teradata database and Operations can see status.

© Teradata 2011

BSI Assignment – Do Better Rebooking

Jodice runs a tight ship and expects fast work on all our client cases. Becausewe all fly, it didn’t take long to understand the problem (always the right starting point!). And all the BSI staff fly a lot so we “know” how frustrating it is – for passengers and for the airlines, when things go wrong.

© Teradata 2011

CRMCRM

Corp LAN

AirLondon’s System Archtiecture

Teradata Production2 Nodes 5600H

Teradata Production2 Nodes 5600H

Dev – 1 Node 5450H Test – 1 Node 4550H

Dev – 1 Node 5450H Test – 1 Node 4550H

ReportingReporting

5TB

Business Analytics Business Analytics Teradata

Relationship Manager

Teradata IWI and Value Analyzer

© Teradata 2011

4 Impacted Passengers Missing LHR Flight

AirLondon uses security cameras at check-ins, so we just loaded the camera shotsand grabbed a frame to store with the Passenger record. It’s just a “blob” of bits to Teradata – but sometimes it’s nice to see who you are helping.

© Teradata 2011

IMPACTED PASSENGERS – Basic Info

Matt mocked up this table, reformatting info from the Customer FF and Bookings tables

© Teradata 2011

Revenues, Profitability and Frequency Scoring

We calculated the information in this table using some predictive modeling capabilitiesfrom SAS, along with Teradata Value Analyzer to compute profit once we knew historical revenueActuals as well as costs. We also pulled info from the future Bookings table to compute Freq Score

© Teradata 2011

Channel Use

Chi and Matt loaded Channel information into Teradata, including records for each interaction and each booking. We then summarized that into a rating (but we could have put a 0-100 score). Notes come from the bailout analysis part of the IWI web tool as well loading some Siebel call notes.

© Teradata 2011

Matt Mocked Up the Frontline Screen for Agents

• One powerful thing we like to do for clients to help move them from traditional “back-office” uses of Teradata to the newer “front-line” or active uses of data

• This helps make front-line people – like gate agents (or ultimately, even the flight attendants) look “smarter”. They can apply the human touch to help AirLondon beat their competitors on customer service.

• So the next shot shows a revised screen shot that we want to provide to gate agents so they know exactly what is going on with each customer.

• The goal is all helpful context on one screen, including history and value, but more importantly What’s Happening Right Now and What Will Happen

• For Lana, we added info about Checkin (the baby) and the real-time Rome bag status

© Teradata 2011

Screen for Passenger Agents - LANA History and Real-Time Factors

© Teradata 2011

A Little Bit on the Screens

• Lots of the BI vendors have screen builders, so we just used one of those

• Screens are made up of Portlets, and the screen rendering engine makes web service callouts using Service Oriented Architecture (SOA) middleware like Tibco (or WebSphere from IBM, or Fusion from Oracle, or Netweaver from SAP, or Microsoft) – doesn’t really matter to Teradata – there are lots of front-ends

• By the time we see it, it’s just a Web Service request that gets unwrapped and SQL comes our way.

• The results go back (often wrapped in XML for interoperability purposes), then unwrapped and displayed by the screen manager

© Teradata 2011

Steffi’s Not Having A Good Day

We put this Call information on two screens – here you can see it on the Channel view screen,and that negative comment could be run through Attensity to get a sentiment score that we could add to the mix of rebooking factors. We also put it on Steffi’s individual screen for agents.

© Teradata 2011

Creative Idea – Interact With Passengers

This is a mockup that Matt came up with. The idea is to connect Teradata directly, again with Portlets, to the Web rendering engine with SOA callouts. We created the ordered list of passengersthat we need to rebook and go through them one at a time, giving them each an option. If they time out by not responding within 3 minutes, then we cancel this screen and move to the next until we have exhausted the possibilities. It’s a simple litle program, but a powerful idea!

© Teradata 2011

Creative Idea

The application paints the possibilities for each customer, and since AirLondon has direct connections with the Flughafen Hotel (and others), we can automate the lookup to see if theyhave availability. We also go ahead and print a letter with the updated gameplan, including the dinner voucher and updated flight itinerary for the next day’s flight, and hand that to him whenhe arrives. Teradata also can send it to him via SMS or e-mail messages if he prefers.

© Teradata 2011

Solution!

• Conrad and Jason were the BSI team’s top two picks

• However, when we interacted with Conrad, he’s tired and would be happy to a take a mid-day flight the next day

• And Jason’s home is in Frankfurt so given the delays he doesn’t want to go on to London and cancels

• So now we have room for both Lana and Steffi (who is still unhappy)

© Teradata 2011

The Better Rebooking Rules Engine

There is really no limitto the number of factors that can be added to theRebooking Engine.

All the factors are modeledwithin the Teradata TravelLogical Data Model

Over a period of time, as next steps, it would be important to compare the various treatments of customers with their subsequent bookingactivities and customer feedback surveys to see what’s working and not.

© Teradata 2011

Takeaways?

• You can use data to make decisions

• The decisions may be complex – with tradeoffs because of capacity - which is why you may need to automate the decisions

• The rules you use will evolve over time, be refined

• You can measure the consequences of decisions – good vs. bad

• You can tie decisions to business goals

© Teradata 2011