The Jagged, Fractured CRM Market I
The recent Siebel numbers weren't happy. The Street explanation
is that the ERP players are grabbing market share. I don't think
this makes much sense.
To explain why I feel that way, I first have to argue that
the way CRM applications work is fundamentally different from the
way ERP applications work, and then show why
this fact has consequences for the market.
That's what I do in this piece.
I'll follow up with a later piece on some recent developments
in the market, including more on the Siebel numbers,
Microsoft, salesforce.com, and JD Edwards.
The CRM Problem
In any enterprise, you would expect that the money spent
on automation would be allocated to the various parts of
the organization in roughly the way the rest of the
expenses are allocated. If sales is 35% of total costs,
then you would expect
that software and hardware and infrastructure
that supports sales would get about 35% of the IT
That doesn't happen. On average, over all industries, back office
operations get a lot more computing buck, relatively speaking,
than the front office. Everybody has a more or less expensive accounting
system, but salespeople still pretty much subsist on their PCs.
That fact is one of the reasons that investors have been excited about
the CRM marketplace for the past 5 years or so, and it's a big reason
why vendors of back-office systems have begun to claim that they
sell CRM software. They see far less automation in the front office
and therefore, untapped
Before getting too excited, though, they ought to ask why the computing
buck gets misallocated in the first place. I think it's because
customer facing activities are not terribly easy to automate. Less
money gets spent, in other words, because there's less that
technology can do.
The recent negative press about CRM projects would certainly seem
to suggest this. (See the Nucleus Research report on Siebel ROI
or the figure thrown around by my old friends at AMR, who said that
only about 15% of CRM projects (whatever that means) have positive
ROI.) But these reports haven't really explained what failure consists
of or why it is so common.
Some of the failure is surely due to the fact that far less is
known about making CRM stuff work than is known about making ERP
stuff work. A whole lot of early-stage SAP implementations didn't
work either. But I think much of the problem is intrinsic. It seems
to me that CRM activities are much harder to model than ERP activities.
Consider these facts:
- An action (called a transaction) in an ERP system is heavily
ramified; that is, it potentially affects virtually every other
part of the system. Shipping a sales order, for instance
decrements inventory, generates an invoice, reduces demand, creates
a commission, puts a load on a truck, etc. Most actions recorded
in a CRM system will never have any effect on the rest of the
- Processes managed in ERP systems are usually governed by well-known
and rigorous standards. MRP or cycle counting or accounting for
inventory are all pretty rigorous, quantitative processes.
- The government requires you to track and report on most of the
processes modeled in an ERP system.
- Most processes in an ERP system involve a document (usually
called an order) that goes through multiple states.
By contrast, customer-facing activities tend to be far more open,
episodic, and inconclusive than the back-office activities that
are automated. And this has consequences for the automation.
- Most actions recorded in a CRM application have no effects on
anything else in the system.
- Most activities recorded or modeled in a CRM system are not
governed by well-known and rigorous standards.
- Quantitative analysis of CRM related activities is far less standardized
and requires far more intelligence than, say, creating a standard
- While the core processes in CRM (pursuing an opportunity,
taking a call, doing a service call) are stateful) the states
are nowhere near as standardized as they are in an ERP system. The
user needs to define them.
When you consider these facts, it is easier to se why automation
of CRM gets fewer bucks.
It costs more-modeling the not-standard, intermittent CRM
processes takes more setup and management--and is riskier than automation
of a GL. And the benefits are harder to extract. You don't necessarily,
for instance, get nearly as much leverage out of tracking CRM activity
as you do in an ERP system.
Getting Value from a CRM Application
I'm not saying that CRM systems can't provide benefit. What I am
saying is the following:
- The smaller and more focused the CRM application, the easier
it is to get it in and make it useful (within the limited sphere
where it works).
- Unlike ERP solutions, CRM solutions don't necessarily get better
when their size and coverage increases. With ERP solutions, there is
a clear benefit to having all the activities on a single platform.
With CRM solutions, the costs of creating a maintaining a single platform
can easily overwhelm the benefits.
This is a natural consequence of the facts stated above. If a system
isn't highly ramified, then you can much more easily carve off just
what you need, and there isn't much loss in doing that. If the processes
are not standard, they're harder to manage if only because they are
far more likely to
change frequently. And the bigger the system, the more complicated
it is to report on.
As a practical matter, everyone knows this. HIstorically, the "CRM"
market was always a group of mini-markets (marketing automation,
SFA, call center, service, etc.) with no notion that they should
all be integrated.
Then along came Siebel. Siebel brought a technical platform that
was very well designed, products in all the mini-markets (with more
on the way) and a C-level selling strategy that played on the executives'
lack of knowledge about the technical requirements of CRM systems.
Again as a practical matter, I think Siebel understood what I'm
saying very well indeed. Their product set is in fact a group of point solutions
designed to do specific functions built on a common technical platform. But
they marketed it as a platform, as something that gave you multi-channel, 360 degree
Unfortunately or fortunately, Siebel is now being hoist on its
own petard. The ERP companies are coming along and saying, "We
have a common platform and technology, and it's way broader than
Siebel's. With Siebel, you have to integrate to the back-office
systems, but with us, it's pre-integrated."
If my theory about CRM is right, then this a statement like this
should be viewed very skeptically, because this
back-office integration isn't necessarily such a good thing.
(I realize that I have just sacrificed most of the credibility that I've
built up so painfully over the past five years.)
Integration would really be a great thing if CRM activities really
were stateful and ramified, just like ERP transactions.
Then, each time you reached a new stage
in the sales process, you could feed the information to the back
office. As you close in on a sale, you could even adjust your manufacturing
Unfortunately, most of the time, taking the client out to dinner
makes no difference whatsoever to your manufacturing schedule. So
most of the time, integration that makes a connection between front
and back office isn't all that important.
But without this clear connection between the back-office and the front-office,
issues of integration have to be set against issues of usability
and manageability. And the bigger and more integrated the system,
typically, the less usable and the less manageable.
In ERP systems, for instance, there is something called a customer
file, which contains the addresses of all the companies that have
bought from you. This customer file is usually a mess, because the
wrong names get entered in, etc., and this makes reporting on customers
But believe me, the problems associated with a customer file are
as nothing compared with the problems of managing a contact file,
a file containing the names of everybody you've had contact with
at a company.
And the problem of managing a combination customer and contact file
can be almost the square of the first two. This combo file (which the
ERP companies sell) can be just the thing
you need, in several, important circumstances. But having the two
together always increases the management overhead and complicates
reporting. (Call if you want a long explanation of why.) And it
tends to decrease the usability of the contact file. (Again, call.)
So you have to really need the integration before it's worth while.
What evidence do I have? Well, if I were right, you would see
ERP players coming to market with a platform--integrated customer and contact
files, integrated opporunity and sales order management are the hallmarks in SFA--but
few companies using the full platform, and then only for highly specific reasons. You
might see ERP sales or SFA sales or contact center sales, but not to people who are
buying all three at once and leveraging the integration.
You'd also see a redefinition of CRM by the ERP players who are frustrated
by the first fact. In that way, they can count
things that aren't CRM (like sales orders) and thus lay claim to building market share.
And you'd see a lot of point solutions reeling for a while, but then coming back.
Isn't that exactly what you do see?
If I'm right moreover, rational companies wouldn't typically
look for a platform solution in CRM. Most of the time they would look
for niche customer-facing solutions that solved highly specific problems.
Yes, they might buy Siebel or PeopleSoft call center. But
only if the functionality is designed for their industry,
if the integration that is provided or the integration tools
are really good, and if the
pretty good platform really does reduce the total cost of ownership.
In the past few years, you've seen a number of companies that have
behaved irrationally (in my view) and bought "integration," no question.
But they've paid a lot, and they traded away a lot of usability and manageabilty.
I think this is one reason you're seeing such a chorus of
complaint, argument, and counterargument that CRM doesn't work.
In many of the cases that spawned the complaint, CRM did work. It
solved a local problem quite effectively. But when people tried to
extend the solution and the platform, the problems of usability and
manageability became too great, and the original vision of end-to-end
customer management was abandoned. Did it work? Of course it did. It
provided value. Of course it didn't, too much of it is shelfware.
Talking about the Market
For those of us who watch the marketplace, it's easy and convenient
to say that there is a single market called "CRM." But doing
so incorrectly valorizes the platform and ERP solutions, at the
expense of smaller solutions that often return more value.
I think that if you really want to understand this marketplace,
you just have to go deeper. Why view Pivotal,
Selectica, E.piphany, Click Commerce, and Salesforce.com
as playing in the "same" space, when in fact their products
are completely different, the benefits of implementing them
are different, and the buyers are different? If the CRM buy were
fundamentally a platform buy, it might make some sense, but
if it isn't, then each has to be viewed relative to its own
If Siebel, SAP, or PeopleSoft actually play
in that market, then they, too, are competitors, with some share. But it
is simply misinformed to say that SAP dominates that market when they've
sold a whole lot sales order seats to existing customers.
To be rational about this market, in other words, you have to
recognize that it is highly fragmented and very complicated.
You have to believe that most
buys of customer-facing automation are or shouild be intended to solve
We talk to a lot of companies that have been forced to call
themselves CRM companies, more than in any other category by
far. For 99% of them, in order to figure out whether they can
succeed and provide value, we have to look at very complex questions.
To figure out whether Selectica, to pick an arbitrary example,
is going to do well, we don't try to figure out whether they
are gaining market share vs., say, E.piphany. We ask questions
like, "Is guided selling considered to be a proven marketing
technique in the automotive industry, and if so (I'm not sure),
how much are OEMs likely to invest in guided selling solutions
over the next two years?"
Let me close by returning to Siebel. If I'm right, Siebel's numbers
can't be explained by the "growth in market share" of the ERP players.
That market share was gained only by expanding the definition of the pie
to include areas where Siebel does't compete.
what does explain the numbers?
Let me hazard an answer. Let me start by reminding you that Tom Siebel and the rest
of that group really do understand
exactly what I'm saying above, probably better than I do.
They don't really sell a platform; they sell
hundreds of highly industry-specific products that run on a platform.
I think that whenever you do that, you're going to run into periods
of market rotation, when you've saturated a set of industries and
problems before you've developed the market and products for the
next set of industries and problems. I think, roughly speaking,
that's what's going on there.
But let me go into more detail in the next piece.
To see other recent Short
Takes, click for a listing.