There was an interesting article in the International Herald Tribune / New York Times last week about how airlines are applying more analytics into the area of avoiding overbooking - leading to “bumping” of passengers off the flight. As a business traveller who seems to suffer this on occasions (apparently the solution is - where you can - to check in online as early as possible) I can appreciate any reduction in such deliberate service level reductions by the airlines.

Interestingly the article quotes one of the those people who exploits such poor airline service by deliberately booking on likely overbooked flights, just to volunteer to be bumped (with associated benefits). Perhaps the event-driven approach to solving these overbooking conundrums might be to track the flight bookings by these “Professional Bumpees” - if they are booked on a flight, then the airline needs to take care over overbooking!

Of course, if you starting avoiding bumping the professional bumpees, then they will soon fail to be a good indicator of your overbooking status…

In addition, of course, analytics are only a guideline based on “past events”. What you really want is to embed those “smarts” (knowledge based on experience) with the reactive “smarts” (CEP) that let you respond optimally to new situations - such as weather, aircraft maintenance, or even volcanic events!

Opher Etzion recorded some action items for the Event Processing Technical Society at the EPTS5 symposium last week. One of these was “promotion of EPTS” as the go-to resource for event processing information (hence the irony that these action items are duly  listed on Opher’s blog and not anywhere on the EPTS web site).

One of these was the idea of an ROI Working Group to share and propagate ROI stories (this being a safe activity for vendors to share as customers seeing ROI will not likely be poachable by other vendors!). But how should ROI be measured and reported?

Coincidentally a colleague just mentioned an airline customer who just went live with a new track-and-trace CEP application -  an application with an estimated €2M per year payback.

Some observations here:

• This was an IT project completed during a big downswing in the airline industry, whilst many other conventional IT projects were being cancelled or postponed - but the ROI was compelling enough for this project to survive.
• The application used a distributed rule engine architecture, exploiting TIBCO BusinessEvents‘ agent-based architecture.

Another coincidence: I was just listening to Paul Coby, CIO of BA, talking about the need for lean and agile approaches in the airline industry, at IRMUK’s BPM2009. His message was ‘there are no IT projects, only business projects’ - all with business goals and KPIs to be measured. Model-driven CEP applications like the airline use case above certainly qualify as lean and agile… I wonder if BA is exploiting CEP like its competitors are?