AA Ireland Builds Easy-to-use Predictive Models with New Analytics Platform
Empowered and connected users, progressive and easy change
"In each of the last five years, the industry lost a quarter billion in motor insurance," says Colm Carey, chief analytics officer at AA Ireland. "So, there's a huge push for new data, models, ways to segment and pick profitable customer types—and get a lot more sophisticated. Our goal is to optimize pricing, understand the types of customers we’re bringing, and the types we're trying to attract. We would like to tie that across the business. Marketing will run a campaign, trying to attract a lot of customers, but maybe they're not the right type. If the call center is going to be very busy that week, what's the point of driving a lot of demand there?"
"We wanted to step away from industry standard software and go with something that was powerful and future-proof. In 2016, we had an opportunity to analyze all software. With TIBCO, there were a lot of opportunities around fraud and customer value. Those are extremely difficult to calculate in live environments. TIBCO was the only provider that could simultaneously optimize pricing, provide a deep understanding of customer types and value, and prevent fraud—so we chose TIBCO solutions,” Carey states.
"First, we had to train the analytics team," says Nina Jensen, senior quantitative analyst at AA Ireland. "TIBCO gave us great support. In three months, we had our first model built. That beat expectations. Now we're really up to speed."
"We're using TIBCO Spotfire software for all pricing analytics," says Carey. "Let's say there’s a sales conversion drop-off over five weeks. Rather than someone approaching my department to say, 'Can you dive into this and come back with your findings?' they're going to be able to click on that metric and run a model that shows why that trend occurred. We’re putting analytics in the hands of each business user. If they want to deep dive into the data or do predictive analysis, they have the opportunity."
Carey continues, "Insurance companies have used the same software for 20 years. Why change? It's a huge risk. You would have to train staff or hire people who already know the standard software. Well, I can hire people with no coding or math background, just very bright PhDs, and within three months, they're releasing models. With the old software, it would be a year. TIBCO is easy to learn. You give the capability to build models to business people who understand the business, modeling, and math better than IT. You don't sit in an IT queue for a year and a half. You build a model yourself and generate a lot of revenue for the company. It’s that power. You can build it, and you have the computational power to do things like fraud identification and embedded customer value and then update those models in live environments.
"Insurance has always had predictive models, but we would build something, and in three months, update it. Using TIBCO solutions, data comes in and goes out to models seamlessly without disruption, basically providing real-time predictability. Rather than just predicting, it's 'If I increase or decrease discounts, what's the uplift in volume and profitability?'
"You can understand the total opportunities and risks in the market and make informed decisions. And then, you can ask: What should I do differently? How should my pricing change? How should I facilitate that in the call center? You can understand it all—plus segmentation, fraud modeling, and underwriter profit. We’re going to use it for long-term predictability of call center capacity and CRM modeling, campaigns, and return of investment from them, and how to price a product.
"We're getting software from our parent company—new marketing, call center, and CRM. Should that not all sit in our TIBCO platform? We have all the data and models there and a highly skilled, fully trained team. To optimize revenue and profits, all those systems need to talk to each other. It will definitely separate us from our competitors."