PROJECT M
 
DJ Patil
Ralf Schneider
Sean Gourley

Conference Call: Drilling for insights

Big data is without doubt a promising resource, but the guidelines for extracting and refining it still have to be written. Handled with care, data security and data privacy will become a competitive advantage for companies.

Conference Call: Drilling for insights

Big data is without doubt a promising resource, but the guidelines for extracting and refining it still have to be written. Handled with care, data security and data privacy will become a competitive advantage for companies.


Ralf Schneider

Gentlemen, big data has become one of the world’s most important resources. In a way, it resembles oil: it needs to be extracted, refined and used responsibly to make full use of its value. What is your take on the current discussion?

DJ Patil

Behavioral economist Dan Ariely once compared big data to teenage sex: everyone talks about it, nobody really knows how to do it, everyone thinks everyone else is doing it, so everyone claims they are doing it. “Do you have a big data strategy?” is now the hip question to ask at cocktail parties, replacing “Do you have a social media strategy?” However, its opportunities, as well as its risks, have to be taken seriously.

Sean Gourley

There is a lot of hot air in the discussion, but you can’t ignore it. The analysis of information created by our everyday use of computers very likely alters the way we live. One economic consequence is that granular information about individual preferences can improve pricing structures and increase cost efficiency across all sectors by 5% to 10%.

Ralf Schneider

The sheer size of data makes the topic impossible to ignore, particularly for insurers. We expect the amount of data to increase by factor 40 over the next 10 years. At the same time, computing power continues to grow exponentially, allowing us to refine the data in a sensible and responsible manner. But let me briefly define big data for the purpose of this conversation as proprietary information generated and owned by an agent – for example, a company. In a second step, the information is analyzed, possibly in conjunction with additional data gathered from outside sources.

DJ Patil

That works for me. And you’re right, Sean, there’s more than just talk. Agriculture company Monsanto, for instance, just paid approximately a billion dollars to acquire Climate Corporation, a company that analyzes data to provide crop insurance to farmers. On the other hand, there are clear risks to big data, and we’re struggling to define good practice.

DJ Patil

What is best practice at your company, Ralf?

Ralf Schneider

At Allianz, we do less than we could. We now have the ability to analyze multitudes of unstructured, unrelated texts and figures from various sources with a varying degree of accuracy. Its quantity allows us to accept a degree of imprecision, and the advantage of big data is that we are now in a position to extract meaning from such messy information. We also have more accurate predictors for the risks we insure. The danger is to rely too much on correlations which say nothing about actual causality.

Sean Gourley

The predictors may become as detailed as how a person’s education and driving style affect the premium of his car insurance.

Ralf Schneider

True, but the concept of insurance can only function if risk is distributed across various members of a group. For insurers, information analysis has always been at the core of understanding the risks they accept. So while this is not new, the basis for finding new patterns has expanded. However, neither the patterns nor the data are personalized. What we can do is show how a group with similar age, gender and education behaves – to the benefit of our clients. Big data enhances our ability to identify and respond to individual customer’s needs.

Sean Gourley

Can you give us an example?

Ralf Schneider

Sure. Take liability insurance fraud. These cases often share common patterns: they take place in the home of the insured, with visitors; most objects reported as broken are alike; and the relationship between the insured and the culprit is similar, too. If a claim seems suspicious, we look for aberrations in the patterns to confirm our initial suspicion.

Ralf Schneider

Do you gentlemen have any advice on how insurers should handle big data?

DJ Patil

I recommend that larger corporations in general do a data review just like they do a risk review before they roll out a major project.

Ralf Schneider

What exactly do you mean?

DJ Patil

Just because we can with data, doesn’t mean we should. Every company should have an internal process of data checks and balances in place that allows them to make use of big data in a responsible manner. Similar to physicians, big data users need to make efforts to ensure they are acting in an ethical manner.

Ralf Schneider

Absolutely. Increasing opportunities also bring growing responsibilities. The question for us is not “What is the legal maximum?” but “Do we want to do everything that is legal?” My answer is no, and I am convinced that data privacy and data security will eventually become a competitive advantage. In the meantime, we’re focusing on building an appropriate IT infrastructure throughout our global organization to safeguard our clients’ data.

Sean Gourley

I think it comes down to money and transparency. If a company makes money thanks to their customers’ data, they ought to be up front about it and willing to share some of that profit with those who made it possible in the first place: the people generating the data. In the long run, this will increase clients’ trust, counter Big Brotherlike concerns and enable us all to make the most of big data.

DJ Patil

A data-oriented company culture is essential. The best data-driven companies start with what we call “silenced sustained data reading” where they take 15 minutes just to look at the data. Once that is complete, then they can dig into the questions. It’s about being intellectually honest as a team. Data is a team sport, and it will play out its benefits when it’s democratized – that is, when accessed by various entities within a company.

Ralf Schneider

So far, we spoke about technology and its potential. What role will humans play?

DJ Patil

I think this is a false dichotomy. Humans have one great advantage over technology: intuition. The best data scientist is one who brings intuition and information together and moves effortlessly between both areas.

Ralf Schneider

That’s for those of us who are prepared to work with data. What about the rest?

DJ Patil

Well, it is the data scientist’s job to make data accessible. From then on, almost everyone can be educated to deal with data. And we have already done this successfully with technology. I can’t think of a single three-year old who is incapable of using an iPad.

Sean Gourley

Owning the information and the computing power is not enough. That’s like having a Ferrari in the garage but no idea how to drive it. To get the most out of large quantities of data, human expertise – with the support of algorithms – needs to structure it in a meaningful way. The interpretation of data and its patterns will remain a human task.

Ralf Schneider

How do you see the future of big data evolving, Sean?

Sean Gourley

I see three major trends. First, I expect people to demand more value back if their data is used to generate profits. Second, we will see more visualization, so interfacing with big data will become easier. Lastly, there will be a move from prediction to persuasion engines. While predictions like that of Nate Silver’s 2012 US election outcome are impressive, they forecast the future based on past actions. More interestingly, we could arrive at a choice of future scenarios independent of the past. I tend to think of this as persuasion engines, rather than prediction engines.

Ralf Schneider

That is very futuristic indeed. Thank you very much for your time and insights, Gentlemen.

Comments