Information Age (IA): With 12 million members, 4.2 million call-outs a year and 3,600 patrol vehicles, the AA must face a huge IT management challenge. What kind of infrastructure underpins that scale of operation?
John Seymour (JS): We have a very sophisticated breakdown service deployment system that has been running now for just over five years. Called AAHelp, it manages in real-time the minute-by-minute taking of breakdown calls from members, finding the best resource to send. That would normally be an AA patrol van, but it could also be one of our garage agents.
The system works out which one to send, deals with the deployment, and sends information on the job to the patrolman or agent over our radio network to the terminal in their vehicle. That’s a pretty dynamic set up. Our phone rings, a member has a problem and we have got 35 minutes to do something about it. It’s five to two now. The breakdowns that our patrols will be working on an hour from now haven’t yet happened. That’s the kind of immediacy of operation we are dealing with.
IA: That must generate mountains of data. How do you link that operational information to your business intelligence system for analysis and reporting?
JS: Every night, at 3 o’clock in the morning, the system feeds details of every breakdown that’s been completed in the previous 24 hours, midnight to midnight, into a data warehouse. So when I issued the daily report this morning, I was telling the business what had happened up to midnight last night. What the patrols report back from the terminals in their cabs is what falls out of the bottom of the data warehouse every day.
IA: What level of business analysis do you undertake on that data?
JS: The data warehouse [analysis system], which we call Service Profile, went live on the first of January 2000, but we had been capturing and retaining data from December 1997 onwards. So as soon as we went live we already had over two years historical data to draw on. We are now at the point where we have more than five years in there. That is important because from a business planning point of view the AA wants to look back over a five-year period.
Service Profile, which is based on an Oracle database, is a 300 gigabyte warehouse, with 24 million rows in the largest table. While it was under development, we used a data analysis tool called DecisionSuite from Information Advantage to get at the data. We bought that in about 1996. Back then, it was the best tool available and it served us well in the initial phase.
But technology moves on. Information Advantage was bought by Sterling Software, which in turn was bought by Computer Associates (CA). CA’s strategy was different; Information Advantage wasn’t a mainstream product [to CA] and it slipped down their agenda. The product could no longer do what we wanted, so we decided to start again.
IA: You chose the BusinessObjects reporting and analysis toolset, from the software supplier of the same name, as a replacement for that DecisionSuite front-end. What factors underpinned that decision?
JS: There were two problems we wanted to address. One was that as we collated and captured more data, a number of our reports were beginning to run very slowly. Some reports that we needed, particularly large ones, were taking 48 hours to run.
We also wanted a product that could handle more complex queries, queries that DecisionSuite could not cope with. We were finding our overnight processes were causing conflict, forcing us to take the system down over the weekend. As we are a seven-day a week organisation, taking the thing down isn’t ideal.
IA: Several other tools could have addressed that challenge. So why BusinessObjects?
JS: We did have a look at some [alternatives], but from things that we had heard, from the general market and networking [with other IT managers], we had had very good reports of Business Objects’s technology. It was also being used already in other parts of Centrica [the AA’s parent company]. That gave us a bit of a head start, so we did a proof of concept, a bit like tasting the wine before you decant the whole bottle so you can send it back if you don’t like it. We were happy with the wine. Other departments within Centrica were also looking at using it, so there seemed to be a group consensus building around it.
IA: What else were you trying to achieve with the new business intelligence tool?
JS: Depth of analysis. For example, if one of our members rings and requests service, in the vast majority of cases we will send out an AA patrol, the patrolman repairs it and that’s the end of the story. But on a minority of occasions we need to send out more than one resource, more than one patrol or involve more than one garage. There’s a whole raft of reasons why that may be the case.
We wanted to be able to analyse those occasions when the breakdown required more than one resource. From a business point of view, it’s in our interests to drive down that number. There may be ways of handling the breakdown problem differently. It may be that if we carried additional spare parts or tools on the patrol vehicle or provided extra training to the agent – a whole raft of things – then the problem can be solved by a single resource. We want to know why it is not solved on the first call out.
In data terms, it’s a fairly simple one-to-many structure, but DecisionSuite did not allow us to group things together in that way. BusinessObjects allows us to build up that structure and to look across the whole profile of the breakdown.
IA: What benefits has it delivered to end-users?
JS: Field service managers, who operate teams of between 15 and 25 patrols, can now use BusinessObjects to examine how often an individual patrolman is calling in additional resources to deal with his breakdown job. A field service manager can say, “The figures on this guy are a bit high, can we have a look at why that might be? Can you send me not just the jobs that he has worked on, but the ones of the people who were called in to help him so I can analyse why he is not able to sort out the problem on his own?”.
It might just be the luck of the draw, the kind of jobs that he’s had or there might be some training or technical issue that needs to be addressed. The field managers love that kind of thing because they can really target specific problems.
IA: Solving those kinds of problems is clearly valuable. But what kind of return on investment (ROI) have you seen from your deployment of the BusinessObjects product?
JS: We estimate an overall ROI of 100% in one year. It is a bit finger ‘in the air’ but by estimating how much time people have saved, we have identified potential savings of more than £1 million.