Back in September, SAP announced the acquisition of KXEN, an advanced analytics Screen Shot 2014-02-11 at 9.09.59 AM.pngcompany that focuses on the predictive space. I recently wrapped up a mini-tour of Asia-Pacific Japan with Shekhar Iyer, our Global Vice President for Business Intelligence, Visualizations, and Predictive to talk to analysts, partners and customers in Singapore, India and Japan about the acquisition and what it means for SAP and more importantly, for our customers. Here’s the story.

It’s an interesting story, different than what analysts were expecting. The first question they asked when we announced the deal was, “Does this mean you’re going to go head-to-head with SAS and IBM’s SPSS?” These are solutions for data scientists, typically PhDs in math or engineering. These are highly skilled people who can work with massive data sets, uncover hidden correlations and build predictive models. Most companies only have a handful of  them. For example, SAP has about 66,000 employees and ~200 data scientists, fewer than 1% of our total employees. It’s a very specialized niche and competing in that market is not what this acquisition is about.

Bringing predictive analytics to the masses

KXEN’s InfiniteInsight is about bringing predictive analytics to the masses via business analysts, who occupy a less rarefied niche. These are not casual business users, but people who are skilled in using business intelligence and visualization solutions who occupy that middle ground between the data scientist and the business user.

This is huge, because a significant hurdle to getting value from big data is the growing shortage of data scientists. McKinsey estimates that by 2018, the U.S. will be short about 190,000 of them. This is a global shortage, and especially acute in Asia-Pacific Japan because what talent there is primarily ends up working in more mature economies in North America and Europe. When you throw in language barriers—it’s hard to get an Australian or Chinese or Indian mathematician to go and be effective in Japan—the shortage here becomes even worse.

That’s consistent with the single biggest thing I hear from customers who have SPSS or SAS:  These solutions are institutionalized, but there are very few people that can use them. Everybody is looking to get more from their big data investment. That’s where InfiniteInsight comes into play.

Addressing the data science talent shortage

In the data science realm, a lot of time is spent building data sets and then doing modeling. InfiniteInsight addresses talent shortage because it automatically creates the models. That means that we can empower the business to create their own models or we can take the data sets generated from the data scientists and then hand them off to business analysts who can use InfiniteInsight to create and identify the best models.  InfiniteInsight leverages business analysts to help data scientists extend their reach so companies can extend their predictive bandwidth.

InfiniteInsight has simplified the approach to predictive with a number of different solutions, both cloud and on-premise. Because they were involved in the SAP startup program they had access to HANA, so their solutions can leverage HANA, but they also support other databases. It’s great for customers who have SAP, customers who have HANA, and those that don’t have either.

How it works is—and I’m simplifying here–you dump in all of your data, and it returns a series of models. It visualizes the results, so you can actually see diminishing returns on a curve. That points the way to the fastest, most effective model.

Predictive modeling for growth and retention

For example, let’s say you’re a telecommunications company that needs to do churn analysis to see which customers are at high risk for leaving your network and intercept them before they do.  So, you might look for correlations with data usage, monthly bill amounts, roaming minutes, gender, age, or income to identify patterns and find those customers.

Then you can add cost elements to see what it would cost to keep each at-risk customer group through promotions. You can try out different promotions and predict which one will get the best results with each group.

If you’re in banking, you might use InfiniteInsight to do what’s called next likely product analysis: for all those people who’ve purchased a particular product, what should you offer them next?

These are basic business challenges around growth and retention rather than super-heavy data science, but someone’s got to build the models and refresh them as conditions change. InfiniteInsight eliminates the bottleneck at the data scientist level and enables the business to do a lot more predictive modeling, at a much lower level of the business.

Precision retailing

On our tour we met with a retail group that wants to do very detailed customer segmentation to make offers that go way beyond the usual coupons. They want to get to the level of, “You shop at this particular location. You come in on Tuesdays, and you buy these certain things. What can we offer to get you to spend twice as much as you usually do?”

I have an amazing retail demo on my iPad that shows how companies can do this. It shows an actual store with foot traffic in real time, because you can get all that information from Wi-Fi signals or video cameras in the store. You can see who’s in the store and where, and if they’re part of your loyalty program or using your free Wi-Fi, you can see personal information about them. You can also see your stock levels.

Then you can actually model different in-store promotions based on the demographics of your location, what’s in stock, who’s in the store and who’s likely to come in. The models will tell you what impact each promotion would have on sales and on stock and which ones are likely to be most profitable or unprofitable. You can do this in minutes, enabling the store manager to make very timely decisions. SAP can then provide the last mile using our mobile technology to deliver the offer via application or email to shoppers’ mobile devices.

I love this, because I always talk about empowering the business with actionable information in the moment, and that’s exactly what this does. It changes how, when and where decisions are made and cascaded down through the business.

As a store manager, not only am I responsible for a certain amount of revenue, profitability is important. The corporate office might say, “These are the 10 promotions that we are going to stand behind.” Using this solution, the store manager can now determine which one of those 10 will be most profitable for his or her store at any given point in time.

The cool thing is, the business user doesn’t need to know anything about big data or linear regression or neural networks. For them, it’s an app, a little tile on their iPad called “store manager.” They hit the button and they can see all this information and it’s incredibly powerful. That’s where we’re trying to take InfiniteInsight. We’re enabling business analysts to push all this information down to the end consumer so they can make optimal decisions.

Additional use cases

We’ve already identified dozens of additional potential use cases where we can embed InfiniteInsight into other SAP applications, including budgeting and planning, revenue forecasting, all the kinds of things that companies try to do in Excel spreadsheets.

There are also some exciting use cases for looking at sales opportunities in CRM and predicting which ones you’re likely to close based the historical information you have. Companies all have limited resources so they want to go after deals that are likely to happen, or the most profitable deals or whatever fits their strategy.

The customers I’ve talked to are all very excited about this, and we already have a good pipeline. They all understand the value of SAP and SPSS. The challenge is there are not enough people who can use it, so the time to value can be really long. The efficiency and effectiveness that InfiniteInsight brings is absolutely critical to the value equation. I’m definitely excited to start selling it.

This article previously appeared on kurtbilafer.com

To report this post you need to login first.

2 Comments

You must be Logged on to comment or reply to a post.

  1. James Oswald

    There are also tremendous use cases in healthcare (How likely is a given patient to readmit? What are the chances someone will react positively to a given medication?) and mining (How likely is this part to fail? How likely am I to run out of inventory for such-and-such a part?) just to name a few.

    (0) 

Leave a Reply