The Case for i2
i2 has been down so long, that up is about the only direction
left. Could it be sooner, rather than later?
Introduction
Readers of this newsletter might well expect an article with this
title to be very short.
Bear with me. I've been watching i2 since 1993, so I do know why
so many investors and customers are angry and disappointed. Even
so, or perhaps EVEN SO, I think there's a good case to be made for
them.
Briefly, it's this:
- The business model is now very sane.
- The customers like i2.
- The latest release, Version 6, keeps i2 differentiated
from its competitors.
- Most of the bad news is in. Not all, but most.
- The almost certain upturn in the software market will help
i2 a lot, more, perhaps than any other vendor.
- The market for supply chain optimization remains very,
very open.
- Time is on i2's side.
These aren't just mild positives. The customers aren't merely happy;
many are unusually happy. The business model is not just workable,
it's forward looking. If i2 had no past and had just descended from
the sky, we'd be enthusiastic.
I'm going to talk about each of these positives, and some still
substantial negatives, in turn. But before I begin I want to look
back at what went wrong with i2. After all, one can't expect them
to do better if they don't fix the problems that put them where
they are today.
Optimizing the Supply Chain
What went wrong at i2? The usual fall guys are the management,
especially Greg Brady, and the software. "The management oversold
and overhyped the software, and then when the software didn't work,
the customers got angry," would be the standard one-liner.
There's a lot of truth to this, but it's not very explanatory.
Any number of companies fit this description. Why did they oversell?
Were they lying, deluded, or ignorant? What was wrong with the software?
Was it badly designed? Badly put together? The wrong stuff? Poor
quality? Or something else entirely. If you want to know whether
things will improve, you need to answer those questions.
I've never been convinced that the software was bad at all.
Poorly designed? Buggy? Hard to understand? Unscalable?
I've heard people say all these things. But it doesn't make sense
to me, and it certainly doesn't accord with my (somewhat limited)
experience.
In my view, it's not so much that the software was bad as it is
that even good software in this area is much harder to install effectively
than anybody )including a lot of i2 people) thought it would be.
And I think that only now is i2 coming to grips with this.
i2 has many products, of course. What they
all have in common is the idea that
(somewhat) higher mathematics can be very
effective in improving the way people
make and deliver goods.
It's a theoretically sound idea. It ought to work.
But it often doesn't.
Why? As Tom Stoppard says, "It's a mystery."
Take two different
companies and apply the mathematics (install the software)
in both of them. Same industry, same size, same business processes.
In one of them, everything changes for the better.
In the other, zilch.
You do see this in other parts of life. No one can
tell you why one dumb TV comedy becomes a hit and an
apparently identical show fails utterly.
Perhaps more apropos, no one can tell why
some experimental therapy works for one dying person, but doesn't
for another. The meshing of technology and situation is very
complex, and you can't fully control it.
With TV shows or with supply chain software,
there is no end of cheap explanations, of course. Why didn't the software work?
Blame management. (If you're a consultant,
just say, "Not enough top management support," and shake
your head, and everybody thinks you know what you're talking about.)
Blame the software. (I've heard this one often.) Blame the consultants.
(Horrors.) Blame the
culture (at the user or the software company.) Blame Greg Brady.
Again, these explanations apply to too many things to be very
interesting. With supply chain software, I think there are
particular and unusual problems associated with making
the stuff work. I think these problems stem from
the fact that
creating a mathematical
model for complex real-world phenomena is a delicate business.
Getting the model right, understanding how it behaves,
coming to believe in it, and using it correctly are all quite
difficult things.
I'm not alone in this belief. Eli Goldratt, author of The Goal,
calls one of his later, impassioned books about the supply chain,
"Necessary but Not Sufficient," because, to get the supply chain
management right, any number of things (including top management support
and
software that works) are absolutely necessary, but by themselves,
insufficient to guarantee success.
With optimization products, there are particular things
that are necessary, but not sufficient. The model has
to be appropriate to the problem. (Many imes,
this means that the stuff out of the box has to be tweaked.)
The relevant data have
to be available and stable and relatively clean. Decisions
about how to demarcate the problems have to be accurate, otherwise
you end up optimizing local behavior but degrading the performance
of the system as a whole.
In one company I went into (not an i2 customer), it was
necessary (but not sufficient)
for the people there to
understand something about the mathematics of aggregation and
disaggregation, relative to forecast accuracy. Without
that understanding, people were
"adjusting" numbers,
and thereby making their forecasts essentially random. That project
was failing when we came in; eventually, it was shut down.
A former i2 employee once told me, "The thing is, every
one of these installations is different. You can't just
drop it in and go. You get it in and you have to play
with it for a while."
This is not because the software is bad, though several
users I know have made this mistake. Ironically,
the more powerful the mathematical model (software), and the
more complex
the problem, the more delicate and difficult the installation.
With products that are less powerful,
the installation is less complex and is more
likely to "work." But less likely
to realize potential benefits.
So, if bad software didn't cause i2's, shall we say,
rapid contraction, what was it about management? Surely
it's silly to think they were deluded or mendacious.
Couldn't it be that people like Greg Brady or Sanjiv failed
to appreciate the magnitude or nature of the
problems that getting this particular software in
poses? They therefore failed to
deploy the resources of the company in the way that
was going to address those problems most effectively.
Was this failure culpable? Well, whenever you promise
your customers a radical
transformation of their business and total benefits amounting to
the GDP of most island nations, you're running a risk.
But I will say that
the prevailing wisdom back then was that getting benefit
from
this software was far easier than it turned
out to be.
Whatever the faults of yesteryear, i2 is now in the position
of the car company that discovered it is actually in the livery business.
For its customers,
you don't
sell cars; you sell cars, drivers, garages, and repairs. This
might (and did, I believe) happen in India many years ago where the combination
of cars, sahibs, and bad roads didn't work unless you supplied everything
that was needed to make the customers happy.
In a livery business, you have to create an infrastructure that
surrounds the delivery of the car that takes care of delivering
needed services. This is what i2 has been doing
for the last two years. But since their investors thought they
were in the car business (and still do), i2 has to
make sure that infrastructure is
cheap, relative to the cost of the car, even though they
also have to be very much more
responsive to customers than a car company can afford to be.
At bottom, the case for i2 is that they are doing
this. They've got a new business model, which allows them
to address the problems of getting supply chain software in
far more effectively, and they've managed to persuade
many of their customers that this is the right
way of doing things.
i2's New Business Model
Four years ago, i2 was a Dallas-based software company that
made CD's and partnered with companies for services. Today,
i2 is a Dallas-based deliverer of expertise, based on software
and services that are made in India. With this structure,
they can focus on delivering a solution, recognizing that it
will have to be tweaked, because they have solution centers
in India that can do what had been very high-cost, high-risk
stuff, everything from cleanse the data to set up
a replica of the customer's system so that a full test can be made.
What stays in the Western countries is the expertise that needs
to be delivered directly to customers. Even that is in many cases
Indian-based. (Tata, for instance, is an i2 partner with solution
delivery managers in the United States.)
With this structure, it is now possible for i2 to take on
projects that require significant amounts of adjustment (custom
software, data cleansing, integration, even product design) on
site.
That i2 is in fact doing this was made clear in
i2's restatement of earnings. The company now reports
contract revenue, revenue from
software sales where significant work is required to make
the software perform as required, as a separate category.
Some contract revenue is actually a holdover from the overoptimistic
sales process of a few years ago.
But the plan, so far as I can
tell, is to continue doing
this, for the obvious reason that this has become part of
i2's business.
Yes, this means that the mix of service revenue is higher
than it used to be, and yes, it means that there's less
deferred revenue and the revenue is contingent. It also can
mean that the sales cycle is more complex (because you're
selling a custom solution). But if that's what selling
this software effectively takes, that's what it takes.
i2 is actually becoming what I called in the last
issue a "custom solutions" company, a company that sells
best-in-breed software that is differentiated by the company's ability
to deliver a solution based on the software. These companies,
can be caught on a treadmill as the ERP companies race
to fill in their feature set. But in i2's case, if the fundamental
problems are as hard as I think they are, this treadmill problem
is nowhere near as serious as it is for others, because what
really differentiates them is their expertise.
Customers Like i2
Almost two years ago, Sam Nakane came into i2 speaking
a heavily accented English and promising a Japanese approach
to customer relations. I can't say I believed it at the time,
but I was wrong.
Inside and outside the company, everyone says that Sam has
had a real effect. In my conversations with customers at the
user conference (in May, before the restatement), it was clear
that they had felt the effect, and it was also clear that they
were staying (intensely) loyal because of it.
One customer told me, for instance, that Sam and Sanjiv
were on the phone with him the day before the stock was
delisted, assuring him that nothing would change for their
company. He appreciated it.
Let me emphasize that this tone is unusual. A year ago or two years ago,
even companies that
had done really well with i2 would describe the experience
as a trial by fire, successful, but
not necessarily pleasant.
The customers I've been talking to were downright enthusiastic.
Version 6 Keeps i2 Differentiated
Because i2 acquired much of its technology and also
because the kinds of problems that are amenable to
mathematical treatment are highly various, i2 Version 6
is more a collection
of products than a single product.
I have gone through a fairly detailed description of most of
what is new in Version 6, but not all. I'm impressed.
Features are a peculiar thing in supply chain software. They
often sound trivial, but with this stuff sometimes even slight improvements can
mean that a certain kind of problem is now within reach.
There are many of these new features in the new version.
I'll give just one example, though there are literally
hundreds. In certain industries, it's fairly common to
have alternate (called "substitute") parts that can replace
standard parts if need be. If you're out of the standard part,
you put in the substitute. Usually the substitutes cost more
or are inadequate in some other way. But if necessary, you use them.
The existence of substitutes makes scheduling production very complicated,
because you often have the choice of paying more and making the
product sooner or waiting for the regular part.
In the industries where this is common (automotive, electronics,
medical devices), deciding whether
and when to use substitutes is mostly a manual process.
Standard scheduling
solutions, even optimized scheduling solutions, can't model
the use of substitutes accurately. (Effectively, the availability
of substitutes means that you can have longer production runs
if you want them.)
Version 6 has a large number of new features
that make it easier to work with substitutes.
Now, the typical CFO, even in one of these industries
is unlikely to understand what that means and is therefore
likely to regard Oracle or SAP's advanced planning as equivalent
to i2's. But people who are working with schedules on a daily
basis will typically
know that these capabilities give i2 a material differentiation.
It would be wearying in the extreme to go over all the i2 products
in all the industries it serves in the same level of detail. In
my review, I did not at all find that i2 uniformly had an edge over
the ERP players or other best-in-breed. But I could see many places
where the enablers that were in Version 6 would allow a project
at a particular company to go forward, where without those enablers
it would make no sense.
Bear in mind that i2 Version 6 is not really going to spark
a classic upgrade cycle. If you're using one part of the
product line to do retail forecasting, you don't care at all
about the addition of substitution capability in Factory Planner.
Recognize too that if it's working for you now, you are not going to need
what was added. But it will allow present and new customers to
start new projects, which may have been on hold before.
If there is any push to an upgrade, it will be due to the
improvements in the user interface (more web-based) and the gradual
move to a web services approach to the program interfaces.
The Bad News is (Mostly) In
The total weight of bad news in the past two years ought to have
been crushing for i2. A precipitous fall-off in revenues. Massive
restructuring of the company. Delisting.
A restatement of earnings that at the very least suggests that
i2 was making a lot of promises that it couldn't keep with the
software it was selling. It's pretty bad. But it's been dealt with.
There are still three things that weigh on i2's prospects. First of
all, there's a continuing SEC investigation. At worst, current management
could be charged. At best, the SEC finds that the past problems with reporting
have been dealt with. I certainly don't know anything more. Second of all,
there are $350 million in debentures due in 2006 (and potentially
callable sooner). i2 has the cash (barely) to deal with this, but it's
still in the position of a homeowner with a balloon
payment coming up. It would be good if it could generate more cash. Third,
there is the .PK after ITWO's ticker symbol, which reduces the size
of the market since many investment managers are not allowed to
trade these stocks.
ITWO's stock has been trading up in anticipation of its earnings
announcement next week. I know very little about what that announcement
will be; I expect it will be moderately positive. But one effect of
positive news will be to put these negatives look less bad. After all,
you still would have tripled your money in the past six months, even
if the stock has a .PK after it.
The Upturn in the Market
Some of this recent interest in a stock that produced
an almost visceral negative reaction in some of my clients
even a few months ago is surely due to a simple fact. There
is indeed an upturn in software sales. And that upturn
will help those
companies most for which people have the lowest expectations.
This upturn is partly fueled by government sales (an area
that does offer some opportunity for i2) and partly by the fact
that past software investments are beginning to be
fully depreciated, which frees up budgets.
If my analysis of i2's new approach the market is at all correct,
then they should be particularly helped by this latter fact.
When large companies use this found money to solve strategic problems
with identifiable returns, projects that might involve i2 will come up.
As i2 becomes more and more able to show even skeptical teams that
they are now capable of delivering a full solution with substantial
returns, they will get a larger and larger piece of that pie.
The Market Is Underpenetrated
These projects do exist. Sanjiv Sidhu keeps on saying that only
2% or 3% of the potential market for these kinds of applications
has been penetrated. You know what? I think he's roughly right.
Most companies manage supply or demand
across the supply chain in roughly the same way they always have:
at most, using MRP or DRP to do a time-phased matching of
estimated supply and demand.
In all the areas where i2 has
software, the software offers a theoretically better way. As noted above,
the theory has not yet been proven by experience and there are certainly
skeptics who believe that the theory is just wrong. (Those people buy
the ERP product.)
But that leaves a lot of opportunities. Remember, many large
companies do in fact buy the theory. They recognize that
i2 has something that can help them
address their problems.
i2's current marketing pitch is very much geared to encouraging this
belief. Instead of selling the whole product suite with a promise
that eventually, the entire supply chain will be optimized, they are
now selling individual solutions, with the promise that this purchase
will give the buyer leverage on the next one. The company is the
same, the products are integrated and work in similar ways.
It is an acceptance of the fact that the mathematically optimized,
fully controlled supply chain is much farther off than had been
thought, but it is also a way of providing people with a way
of getting there eventually.
Time Is on i2's Side
Institutional memories are much shorter than people's memories.
Yes, investment bankers and sell-side analysts were burned. Yes,
companies were burned. (Or at least, as pointed out above, companies
bought software that didn't produce the benefits.) But those
people move on. Those projects get written off and new managers
charged with making something happen come in.
Eventually, the SEC will finish its investigation. Eventually,
the note will get bought out or written off. Eventually, i2 will
be re-listed.
One of the side effects of i2's restatement is that revenue has
been shifted forward into this reporting year and next, which
gives i2 a cushion and also makes the fall-off of the last few
years seem less precipitous (at least to those with short
memories). One effect of this, however, is that it is almost
impossible to compare post-restatement results with pre-restatement
results.
Over time, though, even that negative goes away. The current,
somewhat more accurate way, of reporting revenue that includes
contract revenue becomes the standard. A few years from now, analysts
will be able to report increases or decreases in contract revenue and
pontificate about that.
I'd like to believe that the "eventually" will be darned soon.
But every so often, I have to say, I am reminded that cultures, even
cultures molded by strong-minded Japanese men, are slow to change.
The Ellison cum Brady culture that simply was
i2 a few years ago has not entirely disappeared.
To get a measure of how long it will take, I for one, am
following some promises that the restatement made with great
interest. Essentially, they said that i2 would
tighten procedures and even foster adherence to a code of
ethics. I'd like to see how long that takes. So far, according
to several lower level people inside the company, there haven't
been many Friday afternoon ethics sessions. We shall see.
Conclusion
Conditions have improved at i2, no question. But
it takes time for those improvements to translate into
improved corporate performance. I think there
will be a return to profitability and solid revenues.
But I don't know when. I've been optimistic ever since
May, but so far (and this is a week before the
September earnings report), license sales have been poor.
If the September earnings do contain some pleasant surprises, well
and good, but I have no information that that's the case. If so,
the success has certainly be gained by a lot of hard work and is
therefore deserved. And if not now, then soon, I think.
To see other recent Short
Take, click here for a listing.
|
