IBM announced recently that it would purchase statistics vendor SPSS for approximately $1.2 billion. Various commentators have been saying this is a very big deal:
- Bob Evans of InformationWeek: “I think IBM’s acquisition of SPSS will mark a seminal moment in that company’s evolution”
- Merv Adrian: “With one stroke, IBM has signalled that it believes itself ready to redraw the BI map…. Why does it matter so much? Simple: Predictive Analytics is what’s next in BI”
- And, of course, IBM’s own press release: “With this acquisition, we are extending our capabilities around a new level of analytics that not only provides clients with greater insight — but true foresight,” said Ambuj Goyal, general manager, Information Management.”
I personally* think it’s probably a good move for SPSS and IBM, but I’m a little more skeptical. Yes, I believe that predictive analytic technology can provide great value, and that it’s an important part of the BI future. But here’s three reasons why the reality – inevitably – can’t quite live up the expectations.
Predictive analytics isn’t a miracle new technology
As I wrote back in 2007 (so no, this isn’t just sour grapes) in a post called Who Really Wants Predictive Analytics:
“…who wants to “drive looking out of the rear view mirror”?… Despite the obvious uses of this type of predictive analytics in organizations… it has not been implemented widely.”
The post generated a comment from Malcolm Ryder:
“So who wants predictive analytics? … *everyone* wants predictive analytics; a subset of everyone wants to pay for it; a subset of that payers’ group is willing to trust it; and a subset of that trusting group is willing to depend on it. And the difference between what makes predictive analytics successfully useful versus successfully marketable is a gap just as large as the miniscule percent of dependents versus the whopping percentage of enthusiasts…”
Predictive analytics, statistical modeling, data mining, etc. are extremely valuable technologies, and have had great business success in a wide variety of applications including fraud detection, cross-selling, and direct marketing.
But they’ve been around for decades, and readily available to anybody who wanted to use them. Even IBM has already sold the technology in the form of Intelligent Miner for many years (here’s a fun press release from 10 years ago, IBM Launches Major e-Business Initiative Emphasizing Business Intelligence).
And there are clear limitations to anything that claims to provide “true foresight”. Bill Inmon talked about the difficulties of Predicting the Future last month on the BeyeNetwork, and the Wall Street Journal published an article a couple of years ago called Now, It’s Business By Data, but Numbers Still Can’t Tell Future pointing out that even the most detailed statistical analysis has limitations since conditions may change, rendering the analysis misleading, and that while analysis can lead to incremental improvement in current markets, it can’t provide the creativity required to take the next big leap and adapt to tomorrow’s markets.
So why so much professed excitement now? It’s true that the advanced analytics market is growing faster than the rest of BI, and that IBM’s acquisition might accelerate it. But it’s also growing off a much smaller base, and according to IDC, the current growth differential isn’t that large: 12.1% growth for last year compared to the rest of BI growing at 10.3%.
If those rates stay the same, the market won’t be too shaken up in the near future. To put things in perspective, the advanced analytics makes up currently only 19.5% of the market, and in 10 years time, at the growth rates above, it would be 22.2% of the market. This is hardly “redrawing the map”.
As Neil Raden puts it in his blog post: SPSS Is Not the Story; IBM’s Vision for Analytics Is. IBM’s vision for BI started many years ago with the notion of “information warehouses” (although Digital Equipment Corporation was first with the idea, according to Rick Sherman), and took a big leap forward with the purchase of Cognos. SPSS is just an add-on to that strategy, not a “seminal moment”.
Predictive should be aligned with business applications
Predictive analytics requires a lot of expertise to implement and interpret, and IBM intends to provide services to help organizations do this.
But even more so than other forms of business intelligence, the easiest way to make this powerful technology available to a wider audience is to embed predefined advanced analytics deeply into a business processes. This has been the approach of the advanced analytics market leader, SAS).
I helped launch a “data mining for the masses” product called BusinessObjects Data Miner many years ago. The product never took off as a standalone offer and was eventually canned. But I still believe that there’s a great opportunity to use advanced analytic technologies with traditional BI, to augment the analysis skills of users – and all of these techniques are much easier if you already have the business context defined:
- Automatically detecting outliers in reports
- Detecting trends and trend inflection points
- Automatic clustering – for example, rather than arbitrarily creating age ranges like 10-20, 20-30, etc., algorithms can help determine the most appropriate ranges based on the data clusters (e.g. 12-16, 16-21, etc.)
- Influencing variables – what other values might be driving the observed trend? e.g. which benefits seem to be most associated with employee satisfaction? (but note that “correlation is not causation”)
I predict that over the long term, organizations will tend to purchase this technology from an application vendor like SAP or Oracle rather than a platform-and-services vendor like IBM.
Predictive analytics is still a fragmented market
Some of the coverage implies that IBM scooped their competitors. But it’s the nature of the IT business that there’s a more or less permanent discussion about possible acquisitions, and any company up for sale naturally tries to get a few bidders. So I assume that the other large vendors in the space decided that the deal wasn’t right for them, in terms of strategy, price and opportunity cost, rather than being caught out by surprise.
It may just be that IBM needed more of a BI boost than the other large vendors — as Mary Hayes Weier put it in InformationWeek:
“IBM and its Cognos division had 10% of the total BI market in 2008 with revenues of $800 million, up 5% since the previous year. Compared with competitors, that’s sub-par performance. SAP, SAS, and Oracle — ranked No. 1, No. 2, and No. 4 — all had double-digit revenue gains, with the total BI market growing 10.6% to reach $7.8 billion, according to an IDC report released in June.”
If it is indeed about embedding powerful technology in applications, rather than trying to grab existing market share, there are other technologies available in the market.
“SAP’s partnership with SPSS is working well and we do not expect IBM’s intended acquisition to have an impact on this relationship. We have a growing number of customers as a result of this agreement and we will continue to partner with IBM and evolve our predictive services portfolio to meet the needs of our customers.”
“[SAP’s] strategy is to provide predictive analysis features for all of us and not just expert statisticians and analysts who have traditionally used more complex tool sets. This requires new and innovative approaches in a market that is still evolving, and as the market leader in Business Intelligence we are committed to delivering the solutions our customers need to have clear line of sight, particularly in the current challenging business environment.
Some examples of our investments in the area of predictive and statistical analysis across our product portfolio include:
- A wide number of methods available for forecasting across multiple applications.
- Analysis Process Designer integrated in SAP Business Warehouse (BW) and our Customer Relationship Management (CRM) application.
- A recent announcement of the intent to acquire SAF AG which offers simulation, analysis and forecasting. Already this solution is enabling our customers to improve their retail forecasting and replenishment. “
But as Boris Evelson of Forrester says,
“'[there are] Plenty of choices out there, here’s one sample short list: Accelrys, Angoss, Applied Predictive Technologies, DataInfoCom, Genalytics, KXEN, Megaputer, Partek, Psydex, ThinkAnalytics, Xeligence.”
SAS is the obvious prize in the market, but Dr. Jim Goodnight, the company’s founder, CEO, and controlling shareholder has so far not shown any interest in selling – and SAS’s out-of-the-mainstream corporate culture and sales model means that any acquisition by a publicly-traded company would probably be a painful process.
It’s clear, however, that IBM has now set the bar for the others, and we will probably see more acquisitions and integrations in the coming year…
- James Governor, Redmonk: IBM Buys SPSS More Quants for a Smarter Planet
- Bob Evans, InformationWeek: Global CIO: IBM’s Game-Changing Plunge Into Predictive Analytics
- Jeff Kelly, SearchDataManagement.com: SAP, SAS respond to IBM’s planned SPSS acquisition
- Neil Raden, Intelligent Enterprise, SPSS Is Not the Story; IBM’s Vision for Analytics Is
- Mary Hayes Weier, InformationWeek: SPSS Gives IBM Advantages, But Can It Execute?
- James Kobielus, Forrester: IBM Goes Deeply Predictive, Announces Acquisition of SPSS
- Boris Evelson, Forrester: The Next Wave of BI Acquisitions?
- Merv Adrian, It’s On: IBM To Acquire SPSS
* Everything on this blog is my personal opinion, but just to emphasize again: these thoughts are my own, not the official or unofficial position of SAP.