Analyzing IBM analytics
When people last week started reading my IBM eBook (available Friday in paperback from Amazon and most distributors -- make Mrs. Cringely happy and send one to all your friends) the tales of IBM customer and employee woe were generally accepted as simple fact but some people had a hard time with my assertion that IBM analytics will probably not be successful (I said IBM is already too late to that party).
One especially informed reader hit me pretty hard on the topic and I think our conversation is worth repeating here. He’s asked to remain anonymous but I assure you he’s in a position to know.
Reader: The only quibble I have is the point you made about the analytics opportunity, where you mentioned that only about one percent of IBM’s perspective customers will care about it. I respectfully (but wholeheartedly) disagree.
I attended the Gartner annual Supply Chain Executive Summit in Phoenic two weeks ago and there were two overwhelmingly common themes presented by the keynote speakers (Chief Supply Chain Officers at companies including Colgate-Palmolive, 3M, Schneider Electric, Caterpillar and Land-o-Lakes:
Talent Management (including skills, organizational structure and capabilities) in supply chain and operations will heavily shift toward data scientists and modelers; and "analytics is being driven by the blurring of the digital and physical supply chains.
Gartner SC analysts remarked that there are three profound impacts of the Internet of Everything:
1) Business processes will be more autonomous (not just "automated"); as such, multi-attribute advanced analytics, scenario modeling and event detection will be the only practical way to profitably manage ever-more complex supply chains;
2) Business models are increasingly relying on the monetization of data, which will further blur the lines between the physical and digital, as well as hardware, software and services; this has enormous SC implications;
3) Business "moments" -- i.e., sudden, unplanned opportunities and disruptions such Amazon entering the spare parts delivery space and Apple competing with automotive OEMs for the dashboard user experience -- will require the flattening of organizational silos to identify and exploit new ideas.
In short, I think IBM has a HUGE opportunity with analytics. Ironically, IBM’s Global Technology Outlook (GTO) published 5 years ago predicted the rise of the Internet of Everything with startling accuracy. But, like a former colleague of mine from Georgia frequently said, IBM’s response has been "like a pig looking at a wrist watch".
Bob: My point was that by the time IBM has a real analytic product suite ready most customers will already have other suppliers. They’ll have an opportunity only if they can execute on it and for the past few years what they’ve mainly executed is BS. Yes, they can buy ahead of the wave but even that would require someone being willing to sell (there are better acquirers than IBM) and IBM then not screwing-up the acquisition by starving it or meddling. There are simply too many possible points of failure. And Watson? Watson isn’t a platform or even a technology as far as I can tell. They can’t just point Watson at analytics and create a cost-effective offering.
I’d love to be wrong but I’m probably not.
Reader: You are absolutely correct on all points. Watson’s budget, if the rumors I heard are true (and I knew a couple of people on the Jeopardy team so I didn’t doubt them) IBM spent north of $1 Billion to win that game. Marketing and Research shared the cost. For that kind of money, they’d better have won or even Palmisano might not have survived.
IBM does have Cognos (which is actually a suite of products) lots of data scientists (not just in Research but also in SWG) and most of the other bits they need. At this point, they are still in the game because there are no clear leaders yet and customers have more questions than answers. The Analytics team might also be one of the remaining few with the talent to go to market, but just barely, and the clock is ticking.
Bob: Let’s look at IBM and some typical business analytic use cases:
--Analytics in HR is already a well established business. There are many players and many services on the market already. IBM will be competing with established, mature, and experienced companies.
-- Many analytics services have become essentially spam filters. An HR manager recently told me they got 2000 applications in an hour after posting a job opening. Almost all of them were from people who were completely unqualified for the position. People are so desperate for work, they’re applying for anything and everything. Those spam filters are throwing out a lot of good applications. Some people have found the best way to get past the first HR barrier is to copy and paste the job posting into their resume, then edit it to fit in. Colleges are telling their students to do this!
-- Baseball is probably the best example of the use of talent management technology. Baseball’s system works because there are good measurements on most aspects of a players performance. The most successful teams are the ones that balance talent and cost. They don’t go after the best of the best, which are also ultra expensive. They look for the best combination of talent and value. The most successful teams still employ human input on the evaluation and selection of players.
-- Baseball is one extreme where they have thousands of data points on each player. Business on the other hand has very few. To do a good job HR must augment its data collection on candidates.
-- I’ve seen positions where the “system” is solely focused on degree, college, graduation date, and a narrow range of experience. If you’re hiring mindless drones, then this is probably a good system. If you’re hiring people you want to retain more than six months, the system is probably rejecting more good candidates than it is passing. Or someone has an age discrimination suit in their future.
-- Do you remember the news reports of IBM hiring in India where people were filling out the applications for others?
-- Most of the major retailers and several other industries have been doing serious supply chain work for over 20 years. IBM is really, really late to this market.
-- Let’s suppose IBM finds someone who’s been hiding under a rock for the last 10 years and really needs supply chain help. Let’s walk through the process…
+ you map out the current operation
+ you collect data on the current operation
+ you create computer models of the operation and run analysis against it
+ you find ways to improve it
+ you modify the current system, upgrade it, change processes, etc. That could include making big changes to your distribution centers, building new distribution systems, buying new fleets of trucks, etc.
+ you have to make big improvements in the information systems.
+ you have to train the organization on how to do their job better.
+ you have to assemble a new team to operate and manage the new system
+ for the next 5-10 years you must constantly monitor and continuously improve the system
+ you have to constantly revisit the models and run new business conditions through it, then optimize the system for that point in time
-- How much of this can IBM do well?
-- Supply chain is a long term, continuous improvement effort. IBM’s business model is sell, do something, get paid, leave. Under the context of "do something" IBM will stick to requirements that may or may not produce the desired results.
-- If IBM’s supply chain expertise is so good -- why does IBM have so many slow and inefficient internal processes?
Supply Chain and HR
-- Let’s start with the assumption IBM is serious about doing supply chain optimizations and has a good HR analytics system.
-- I happen to know someone with deep experience in the subject, the right academic credentials, etc. That person’s CV has been posted on IBM’s internal system in plain sight for the last 10 years. Industrial engineering degree, Purdue, graduate work, operations research, simulation. They haven’t called him yet.
-- IBM is not looking for people with deep knowledge in an industry to provide the service. This is a great example of how IBM views its people as "resources". If they can sell a service, someone, somewhere with no education or experience should be able to follow a "process" and perform the work.
-- Companies have been doing this stuff for over a decade. They are many well established outfits doing this stuff with a record of producing results. IBM is coming into this market very late.
-- The most successful companies are already doing this stuff.
-- If you were one of the potential customers not yet doing this stuff -- who would you pick to do the job? High priced IBM with no track record or one of a dozen companies who have been doing this well for a decade?
-- It was Gartner that recommended the best way to do big data analytics was to hire a couple people, build your own system, and do it yourself.
Or am I wrong?
Reader: You’re preaching to the choir on the SC topic. That’s where I got my start at IBM. When IBM acquired PwC, they got a fully-dressed SCM practice that could tackle everything from strategy through each function (Plan, Source, Make, Deliver, Return/Customer Service, Product Lifecycle Mgt, Org and Talent, Technology, etc.). So, IBM’s certainly not late to the party, but their entire model, built on billable hours, has devolved into a "butts-in-seats" service rather than the transformative capability it should be. Deciding to completely cut training out of their consulting organization and limit their mobility across clients and projects has further hamstrung their value proposition.
One of the major selling features of IBM’s SC consulting practice was that IBM had one of the highest-performing supply chains in the world (AMR Research once ranked them as high as #3). The year-over-year, incremental improvements in Cost-to-Serve were impressive by any measure. Clients wanted some of that capability for themselves, and IBM took it to the bank. The "We Practice What We Preach" method of sales and marketing was extremely successful at winning SC business -- for awhile. After the mid-2000s, it really became a shell game, as you’ve written about many times now.
Believe it or not … Most companies are still in their analytics infancy. Yes, there are pockets of excellence and some companies (Intel comes to mind, as do a few CPG companies like P&G and PepsiCo) are really, really, good at it. One of these days, have a conversation with Mark Wilkinson at Intel. He’ll blow your mind. However, the majority are are not yet grasping the future of analytics: prescriptive solutions to complex events.
But just as well, you could speak with Paul Giangarra, a Distinguished Engineer at IBM SWG. He has been building analytics engines and applications since the 1970s and is probably the smartest person I’ve ever met (actually, he’s a living, breathing, facsimile of Big Bang Theory’s Sheldon Cooper). At least five years ago he was building proof of concepts for IBM clients using complex event processing and advanced analytics applications that IBM was selling as shrink-wrapped software. He actually built (I saw it myself) -- in a single box -- a space shuttle launch sequence controller as a demo for United Space Alliance at the Kennedy Space Center. It took him and one person doing software configuration (remember, this was shrink-wrap software and stock hardware they were using; no Watson tomfoolery) 90 days to complete the project. I’m an aerospace engineer and these two guys duplicated in three months what it took an army of my fellow propeller heads decades to create. Needless to say, it was more than a bit humbling.
To your point about Gartner’s recommendations … Yes, they do recommend that companies build the teams to do this work themselves. There are several reasons, but one of the biggest is that this capability is essential to innovation and for some companies, their ability to remain competitive. It really would be the equivalent of the Praetorian Guard to let someone else do it.
So, no, I wouldn’t hire IBM, but neither would I hire any of the other big firms. I have good friends who have moved on to the Accentures, McKinseys and Bains of the world. Those companies aren’t really very different and they aren’t much happier. They’re certainly no richer.
Bob: I’d call this a draw. What do you think?