Managing data for future profits

Making sense of your business’ wealth of data can be time-consuming and expensive. CXO caught up with KXEN founder and CEO Roger Haddad to discuss the growing need to execute accurate data mining and predictive modelling.

CXO. When clients seek your assistance with improving their business efficiency, what are their common needs and how do you go about meeting these requirements?

RH. It comes down to using the data to understand and make a better decision. Most of the time this doesn’t happen because businesses don’t have the time or money – it costs £30,000 to £50,000 and around four weeks to build a good analytical model. We are changing this, however, making it now possible for businesses to use the data they already have to make more informed decisions. It is more efficient if you can make better decisions. You look smarter and you are smarter. You have to remember that data is an asset, not an IT program.

CXO. Businesses nowadays hold masses of data on their systems, which can give key indicators about their customers’ habits and needs. How can companies use this wealth of information to their advantage?

RH. Companies use analytics for very few questions. The average human being can cope with only five variables. Five indicators on the customer is your limit. Most of the problems even with an Excel Spreadsheet are that you have 10, 15 or 20 variables and you have to know how to cope with that. If you have 15 variables and you want to put them three by three you have 455 ways to do that.

You get a pile of reports on your desk and you don’t know which one to do first. You pick one and you go there. In fact, that really is where KXEN changed the game with predictive modelling and data mining. In a few minutes KXEN can tell you what is the most important.

The senior architect from IBM, Richard Taylor, said he traditionally thought that data mining was the last thing to do on your data but that he has since discovered that you should actually do it first. It is only with KXEN that this possible. We are changing how to cope with data. You have to have the right tools and products.

CXO. According to experts, the predictive analytics market is expected to develop into a $3 billion industry by 2008. What has fuelled this explosion?

RH. When we started in 1998, data mining and predictive modelling was presented as a dead end – “don’t announce the words ‘data mining’ in your business plan otherwise you will never raise any money!” Things have changed a lot, however, and people are starting to understand that what they would like to do is BPM (business performance management). In fact, what they are really doing though is BPM (business performance measurement). And to move from Measurement to Management you need predictive analytics.

Of course, you need something that is simple to use because you don’t want to have to wait four weeks. We are making this possible. In general, people are fed up of looking in the rear window at what has already happened. They want to look through the front window at what is coming. People perceive a value and decide to exploit their data, rather than allowing it to sit there doing nothing.

CXO. As productivity and profitability affect all staff, how important it is for companies to teach non-technical employees the need for business analytics?

RH. It is key. If you are stuck with an old statistical tool it is impossible because you are not going to make all your staff PhDs in statistics. That is where KXEN comes in. We can turn a business user into a good KXEN user in less than one day. We have seen more and more people without a PhD or any training in statistics but because they know the data and the business, they are able to use the tool to make it happen. This is happening a lot in corporations.

CXO. Customer focus is critical across industry. How does data mining technology play a part in ensuring that the right products and services are targeted at the right audience?

RH. Data mining is actually rarely used. Some businesses need thousands of models and when you ask them how many models they use they will say no more than 10. That’s all. All the remaining jobs are done by rules or are random. We call it ‘carpet bombing’. But things are changing because people understand that it is a mess.

Even for e-mail or SMS text you don’t want to overload your customer with messages because that is spam. People hate to be disturbed by things that have no relevance for them so businesses are realising that analytics and targeting are key.

We are seeing businesses in Europe, USA, Canada and China starting to use our technology to build models on every campaign, be it e-mail, SMS or paper. We are seeing this demand for thousands of models. We call it the ‘modelling factory’. But when you have these thousands of models, what will you do with them? KXEN has an ability to export the models, which allows you to really use the database as a scoring engine.

CXO. How accurate are your predictive and descriptive modelling engines?

RH. Although they are automated, they are pretty accurate. None of our customers has tested KXEN against their old tools and found a big discrepancy. Most of the time we do things in a few minutes that used to take a few weeks. Our results are equal or better. You have to be careful about the quality of a prediction and this depends on your data. KXEN has the ability to use all the data. We have worked on 5000 variables at once and it was a success and came to the result a lot quicker than if it had been handcrafted for weeks. Ultimately, the accuracy depends on the data you input. Sometimes where there is no useful information in your data it is good to know that within a few minutes rather than trying to work on it for weeks and finally come to nothing.

CXO. As the need for firms to make profitable and informed decisions comes increasingly under the spotlight, what is the next growth area for the industry?

RH. Lawrence Summers, President of Harvard University, said: “I suspect that when history is written 200 years from now, it will emerge that something very important happened in human thinking during the time we were alive, and that is that we are becoming rational, analytical and data-driven in a far wider range of activity than we ever have been before.” I feel that this is very appropriate and I believe that KXEN is the one to make this passage from data to knowledge happen. Sometimes I compare that to what Business Objects did for reporting when it started in the early years, putting the reporting power in the hands of business users.

CASE STUDY:
Powergen energises its marketing

We all need it and we all use it. But we’re all surprisingly bad at getting the best deal when we buy it. As a consequence nine out of 10 of us pay over the odds for our gas and electricity.

Powergen, one of Britain’s leading energy suppliers, and part of E.ON, the world’s largest investor-owned utility, is determined to redress this balance with its cost-saving energy packages. But the marketing task of targeting the right prospects with the right products is a uniquely challenging one.

That’s why Powergen has taken a lead in its sector by adopting next generation business analytics from KXEN to finetune the way it targets customers and prospects with the most appropriate energy packages for their needs.

Even early on in its implementation, the results are encouraging according to Powergen’s head of customer relationship management, Mark Perrett: “The first time we used a KXEN generated model to support campaign activity we saw a 20 percent uplift in sales and a direct UK£150,000 saving in mailing costs. Those figures represent an excellent and speedy return on a software investment. More importantly, we also retained 300,000 customer contact opportunities for future campaigns, having been told by KXEN that a successful sale was unlikely to result this time.”

While only recently adopted, KXEN is already on the way to playing an important role at Powergen. The company is keen to maintain marketplace differentiation and it sees accurately targeted marketing as way of demonstrating to customers that it understands their needs better than the competition.

“The market dynamics are interesting,” says Perrett. “Where once the competition was very aggressive with a constant race to acquire customers, now it is maturing and causing all the suppliers to rethink their tactics. We’re in a market where it’s difficult to get competitive intelligence, which makes it all the more important to understand and use customer data. And that really highlights the importance of modelling.”

Resource light
It’s given in traditional modelling that a large team of highly skilled, highly paid analysts is required to get results. The Powergen example proves that with KXEN that does not have to be the case.

“Our experiences are based on a team of only three or four analysts,” says Perrett. “A traditional modelling tool would need a much bigger team than that. I’ve heard of organisations where it’s commonplace to have many 10s of analysts, but here we’ve proved that you can build good marketing propensity models with only three or four people.”

Another thing taken for granted in traditional modelling circles is that it can take as long as several weeks or months to build one model. Here again, Powergen has overturned established thinking.

“With KXEN, it takes minutes to build a model and it can be applied to several million database records in a couple of hours, whereas building models in other solutions and getting the data formatted to suit can take weeks,” says Perrett. “KXEN gives you a choice of outputs and it writes code for you that you can apply straight away to your database.”

Energy focus
Unlike others in the industry Powergen has stayed highly focused on the supply of energy, resisting many of the ancillary add-on services offered by its competitors, hence the company strap line ‘positive energy’. The Powergen brand is part of the retail arm of E.ON UK and has some 8.5 million electricity and gas customer accounts.

Perrett and his colleagues in the Powergen customer insight team are now charged with helping their business colleagues to grow that number. KXEN is being viewed as a key tool in that process by exploiting the customer knowledge held in Powergen’s Oracle customer database.

“KXEN allows us to look at the data we have got without any preconceptions. The problem is when you have a lot of data it’s easy to jump to conclusions but we’d be guessing if we just made simple predictions. With KXEN, we can still go with gut feel but now it's possible to validate that intuition with hard data.” He added: “We got KXEN initially for a very specific purpose but on the strength of our early experiences we're going to invest some time in exploring its other capabilities. There are many abilities in the tool that we're not yet exploiting but look forward to doing so in the future.”