The Ivory Tower
July 7 , 2005

Whither Best in Breed?

Best in breed application companies are poorly understood and severely undervalued--by the investment community, by customers, and sometimes by their own management.

The Opportunity

"Best in breed is dead," Bill McDermott announced at a press conference recently, getting nothing but a murmur of assent from the assembled journalists and analysts.

Like most nakedly self-serving statements, this one is obviously wrong. Best-in-breed application companies by the hundreds are still out there, still selling and still serving.

Little facts like that don't bother canny observers. True, the companies aren't actually dead, they think, but the model is, and thus the life has gone out of them. It's only a matter of time until these zombies go bankrupt or are merged into the Borg, that is, the Big Two or one of the assemblers (Lawson, SSA, or Infor).

What killed the model, which I agree is dead? It isn't SAP, much as they and the canny observers would like to give them credit. No, it was two little words: "Power" and "Point." The traditional best-in-breed model, you see, depends utterly on the best-in-breed company having a two or three-year technical lead on their platform competitors, alias The Borg. But when the large platform companies can use a development tool like PowerPoint to build a competitive product in roughly 20 minutes, the best-in-breed game starts turning into one of those backroom roulette games in a small town. If you do ever start winning, two very large men will appear on either side of you and relieve you of any excess, and a few minutes later, you'll find yourself in a dark alley.

Can anything else happen? I think so. I think best-in-breed companies, new or old, can win and thrive. All they have to do is change the model.

This is the first in a series of pieces on the best-in-breed model, what has happened to it, and how best-in-breed companies can adopt a new model and survive.

No, I'm not just being contrarian, though I enjoy that. I think that we happen to be in one of those interesting times when received ideas (in this case, received ideas about how software companies should operate) are largely outmoded: they literally haven't kept up with the technology. As always, in such times, people who do recognize the changes and adapt can and will do very well.

In this first piece in the series, I'll try to take a fresh look at the best-in-breed business model; in later pieces, I'll talk about individual best-in-breed companies and what they are doing to shift their model.

As I think you'll see, this new model isn't something I made up. It's something these companies are already adopting. But there's no articulated framework for this model; that's what I'm trying to provide here.

I'll begin with what has changed in the best-in-breed landscape, continue with a description of what's wrong with the best-in-breed model, and conclude with some recommendations for best-in-breed companies. In an addendum, I'll talk about some of the things you need to look for, if you are trying to evaluate best-in-breed software or best-in-breed companies.

Just so that everything is out front, let me simply announce that this entire series is based on a rather simple proposition, which involves a fundamental change in thinking:

• A new-model technology company can survive (very well indeed) if it provides value to the customer that is a) available exclusively from that technology company and b) significantly greater than the cost of buying and installing the technology.

Note the occurrence of the word "value." Under the old, crossing-the-chasm model, the aim of a technology company was not to provide value, it was to provide tools that other people could use to generate value.

The great thing about that old model was that you could make pots of money without ever actually taking responsibility for the value that you did (or did not) generate. The bad thing was that it was too easy to soil your own bed. Over the years, so many people sold great-sounding tools that didn't work that people have become suspicious. They realize that the business value they can get from these tools is highly unreliable. (Essentially, this is not because the vendors were charlatans and sleazeballs, it was because gaining value depends on knowing how to use the tool.) Over the past ten years, that unreliability has loomed larger and larger over the application software landscape.

It is now looming so large that it is creating an opportunity for those who can seize it.

Best in Breed: the Old Model

Where did that opportunity come from? Let's look at the old model in more detail.

According to that model, an application company needs two things, an innovative piece of software and a "space" where the use of that software is compelling. If this new software can get to market in the space quickly enough and become the default way of doing things, it will cross the chasm and dominate the space.

This domination is good for buyer and for seller. For the buyer, it means rapid growth and very high margins. (All you have to do is cut a CD.) For the seller, there is a network effect; when everybody uses the software, the cost goes down and there is interoperability.

In the ideal case, the one that venture capitalists dream about, the space is horizontal, so that eventually everybody uses the same technology. The poster children for this is are Microsoft or Oracle, where a piece of technology that everybody needs becomes the property of a (virtually) unique provider.

But the model has grown outmoded, at least for application software. What has happened? Three things. First, we've run out of "spaces" that are big enough. No matter how innovative you are, you're never going to get a Microsoft out of an application that manages caseloads for lawyers with a new and highly innovative technology.

Second, margins are nowhere near as high as people think for mid-size software companies because people ignore the very considerable costs associated with getting applications out into the marketplace. These costs, unfortunately, grow linearly with the number of customers (unlike the cost of developing the software). For mid-size software companies, these costs include the cost of product testing, modification, and release, the cost of a global infrastructure, the cost of lawsuits when customers were oversold, etc., etc.

In addition, for mid-size companies, margins are affected by the fact that they must engage in some kind of service business. You see, when the application company doesn't have the market power to persuade partners to take on this lower-margin service business and provide services effectively, then they have to do it, in order to make sure that the customer gets some kind of use somehow out of the product.

Third, and most important in this day and age, there are no barriers to entry. As noted above, Power Point can construct a good-enough competitor in the time it takes you to say, "Clip Art." Over and over again in the past five years, I've seen the Big Two win deals on little more.

Added together, these are a triple whammy. Best-in-breed companies (most of them) don't have a big enough market. It costs them too much to enter the markets that they can occupy. And if they do succeed beyond expectations, my friend Power Point will take them down in no time.

So should you invest in an old-model best-in-breed company? To quote Shakespeare, "Never, never, never, never, never."

Best in Breed Has Changed

So why are best-in-breed companies surviving at all? Well, underneath all our noses, best-in-breed has changed.

The software is much better.

Take a product like Adage, whose software was catastrophically bad when I reviewed it five years ago--and that's not to mention the quality problems. I talked to customers recently, and most of the problems are fixed. It's no longer just workaround, workaround, workaround and pray that it runs. Adage was not an exception my view; it was typical of produces whose life blood was getting to market fast.

The actual, as opposed to the Power Point gap between what the best-in-breed companies do and what the Big Two do has widened.

Take i2's or Manugistics' 64-bit forecasting products, which are pretty capable and can solve important real-world problems for large corporations. SAP's APO is better, too, but its architecture limits it, and it just doesn't have equivalent capability.

The software is better understood.

Years of dealing with (often) unsatisfied customers have given the companies some (often limited) knowledge of what works and what doesn't with their software. It also has given customers a better feel for what they can expect and what they need to do. This has not necessarily made buyers wiser, but it has reduced the failure rate and reduced some of the costs associated with getting the product out to market and used.

Prices have gone down.

On the one hand, this is bad for best-in-breed companies, since they get less money for each sale. But on the whole, it's positive, because prices at these levels mean that the large majority of their potential customers are no longer put off by what they justly regarded as an extortionate and predatory attitude toward their pocketbook.

All this does not in itself add up to a change in the model. But it does mean that best-in-breed companies are in far better shape than they were a few years ago. If only because most of them are survivors, they have the stability that comes from a more or less happy customer base. Because the software works better, is more stable, and requires less new functionality, their cost of R&D and delivery is down. And because everyone understands the product better, the chances that value will be delivered have improved greatly.

This improvement in best-in-breed players is one reason why the assemblers have appeared on the scene. They've realized that underneath these sometimes still sodden, staggering best-in-breeds, there are some assets, assets that their own management, still nostalgic for the model that brought riches to Larry Ellison, failed to exploit.

Unfortunately, the assemblers (more on them in later pieces this summer) are also caught up in this model. Their theory is that the model fails largely because of size. Best in breed players are too small, they think; the ante (mentioned above) is too high. I think this is wrong, but it at least shows that somebody is thinking about the model.

A New Model for Best in Breed: Deliver Value, not Software

Here and there, you see other companies thinking, too. Ariba inaugurates a service business. i2 focuses on customer success stories. Manhattan and Agile embark on acquisition sprees. Given some breathing space, these companies are consciously deciding not to put all their resources into building more functionality. What they're doing doesn't yet add up to a clear notion about a new model. But it's a move in that direction.

Just to help things out, let me suggest a framework for this new model, which provides some underlying rationale for their efforts, suggests some directions they can go in, and allows you to evaluate their progress.

In a new framework like this, the new-model company must have an elevator story, something that clearly differentiates them from the Borg and gives their customers and investors some idea that what they're up to is coherent. So here's what I think ought to be the underlying pitch for these companies.

We (the best-in-breed company with the new model) deliver value, not software.

Why say this? First of all, it takes you immediately out of competition with the Borg, who are delivering software. And it puts you in a desirable position relative to them. Everybody would rather have value than disks.

This pitch seizes on one of the Borg's intrinsic competitive weaknesses: the aforementioned uncertainty about whether you, the customer, will actually get any value out of those disks.

For those of us who have faced it, this uncertainty is pretty daunting. It always reminds me of how my father felt in the '50s, when he bought an amplifier kit from Heathkit. The box came and it had 1500 parts. Myself, I would have returned it. But he slogged away for six months and actually got some sound out.. Unfortunately, the quality of the sound depended heavily on one's ability to solder connections correctly, something one learns over the course of the project.

With Heathkit, the uncertainty was pretty clear and pretty upfront, and as soon as competitors capitalized on it, Heathking was out of business. With the Borg, their marketing machines and the fact that their competitors have the same model and face the same uncertainty have largely protected them. But the potential vulnerability remains.

If you recognize that vulnerability and differentiate yourself on that basis, by delivering value, not software, you're grabbing a pretty powerful position. Remember, this vulnerability is like Heathkit's the weakness is intrinsic to the software business model.

Not that delivering value is easy. Value, after all, is realized by the customer, not delivered by Federal Express. If somebody promises value to a customer, they need a lot of control over internal corporate processes, control they don't have. So maybe, strictly speaking, you can't deliver value per se. But you can greatly improve the probability the probability that value will be realized.

As long as that's what you focus on.

How to Deliver Value

So how can companies begin to focus on delivering value, not application software? Here are some pretty obvious suggestions.

Reduce the time to value of their software. "Time to value" is a phrase that's begun to appear a lot in the marketing of application companies, and as usual, it has many different meanings. The way I define it, the time to value is the time elapsed between the initial delivery of the software and the time it has returned more value than it cost.

To reduce time to value, you do things like make the software much easier to install, make it easier to use, improve the quality, make it easier to upgrade, and reduce the time it takes to do things like data migrations or integration.

Poster children for reducing time to value are Salesforce.com (more on them in a later piece), the old PeopleSoft TOE (Total Ownership Experience) program, or even Microsoft when it finally made a Windows product that could be installed and run reliably by a human being.

Better time to value isn't just faster installation. It's getting rid of all those impediments (from bad documentation to hard-to-use reporting programs) that make the software hard for the users to manage it or use it.

When you start to reduce time to value by changing your software, you soon discover there is something else you can do:

Lower the process and people barriers to getting value. (A prerequisite for this is to understand what those barriers are, something few software companies do.)

Here's a simple example. One reason that the supply chain optimization packages (like i2 or Manugistics) never worked properly in a lot of companies is that the people who used them didn't believe the numbers that they provided and hence overrode them. The vendors have known this for years, but to this day, they have typically treated it as a user problem.

If they had simply taken some responsibility for the problem, there would have been about six million things they could do to overcome it. Make it easier for the software to run simulations. Make it easier to figure out how a number was generated. Track the effect of overrides and compare it with what would happen if you didn't override. Provide online, point-of-use training or, even better, online, point-of-use expertise. ("Hello, why did the system come up with this number?" "Well, let me look." "It did this and this." "Oh, I see.")

Some of the companies that are beginning to move toward the new model (like Ariba) have thought about these barriers and done something about them. At Ariba, they beefed up their services organization and they picked up FreeMarkets, which provides expertise available almost nowhere else. Naturally, Wall Street punished them for this, and naturally, they've had problems. But it's still the right idea.

In my view, these kindsproblems are largely due to a conceptual weakness in what they're trying to do. If these companies develop service organizations just to get the software in, they are not focusing on value. The companies have to take took the time and make the effort to understand the process and people barriers then provide the service organizations with innovative tools or expertise that could make them far more effective and thus justify higher and more reliable margins.

One reason that software companies don't take responsibility for any of the time-to-value barriers is that doing so looks like pure cost to them. But it wouldn't be if they did the following:

Sell expertise, rather than software, where you can.

This sounds peculiar, but it is actually essential, and it gets at the heart of the new-model best-in-breed value proposition.

You see, for a best-in-breed company to survive, they have to have something proprietary that is exceptionally difficult for the Borg to replicate. Back in the days when data models were hard, you could go to market by saying that you supported X, where the Borg did not. But nowadays, that's virtually impossible. You have to have something that it's difficult for Vinod Patel or Chao Chen to replicate, and that isn't software.

The poster child for this strategy was FreeMarkets, before the market realized they were overcharging. Anybody could build auction software--and did. But you needed expertise to run an auction in, say, some commodity alloy, that would actually generate an excellent price. What FreeMarkets did was sell events. They used their expertise to find the right participants and run it in the right way, plus they supported that expertise with software that was tuned to their way of running auctions.

But isn't this a special case? No, I think it's applicable to almost any area where a best-in-breed company wants to play. Software, you see, is fundamentally something that automates activity. Almost by definition automatable activity is dumb activity, the activity that produces the least value. To make money replacing low value-add workers, you need scale; you need to replace a lot of workers.

The Borg can do this, because they're big and they sell commodity software. But for best-in-breed software, which cannot scale all that much, you need to find some way of generating greater value-add.

People routinely call this service, but it's not service at all. Service is something you're paid for by the hour, where you do something that others could do. The real way to generate value-add is to provide expertise, where you're paid not to do something, but to provide the fruits of the knowledge that you've spent years developing, something your customer can't possibly acquire. Expertise can be as profitable as software is (remember that intellectual property can also be duplicated with very high margins), but it is much harder to replicate than software is.

There are two other really great things about building a business model partly on expertise:

• Expertise, unlike software, needs to be renewed. You can't buy last year's expertise and then use it when you have a new problem. So it makes sense to sell it on a subscription basis, which is a terrific revenue model.
• The Borg simply "can't be in the business of providing expertise" as several senior executives at these companies put it to me. So, while customers can go to the Borg for software, they can't go to them for expertise.

By getting in the business of providing expertise, then, not only do you make the delivery of value more reliable, but you do it in a way that provides a beachhead against your competitors and improves the flexibility and reliability of your business model.

Mine the customer base. Best-in-breed companies must accept the fact that their value proposition is not the same as that of a pure tools play. For them, the kind of growth that is usually expected of software applications is simply inappropriate. They can still be attractive, because they can generate very high and growing margins. But they just can't be Microsoft.

A pure software company, remember, spends a lot on getting new customers quickly. But where there is no chasm to cross, an emphasis on getting new customers can distract you from delivering value.

For a company that delivers value, a dollar spent on delivering that value has three benefits. First, it gets an immediate return, because you're being paid for delivering the value. Second, it solidifies your position with the customer and therefore gives you the opportunity to sell something new to a customer who already knows you.

Third, and most important, it allows you to make a persuasive and differentiated case to new customers who are like the old customer. When you're in the business of delivering value, the natural question for new customers who is used to sleazy software companies is, "What proof do you have?" Demonstrable success in a company that is like their own is what they are looking for and what only the best-in-breed company (new model) can provide.

Evaluating the New Model Best-in-Breed Company

So what does one look for in a new model best-in-breed company. How does one know when a company like that is successful.

First, what is success? Clearly, it's growth that is more reliable, though more moderate than the growth of the pure software companies, but more reliable growth. It is very high margins. It is a very loyal customer base, which promises good future returns. It is no competitors capable of making inroads. It is a well-respected brand. A bar high enough that any of today's best-in-breed players would be happy to leap it.

So then what are indicators that a company is moving in this direction?

• The company understands the value it creates and has evolved a way of helping customers get that value reliably.
• The company uses a value-based or solution selling approach and uses some kind of ROI or expected value selling tool.
• The company will typically sell to business users or the CFO, not to IT.
• The company persuasively differentiates itself from the Borg by citing the value it provides.
• The company can answer the "platform" argument, the argument from IT that says it is better off running a single platform than it is adopting a host of best-in-breed applications.
• The company carefully tracks the use of its software and prides itself on demonstrating that the software is used effectively.
• The company invests in making its customer-facing operations more effective.
• The company is highly focused on specific verticals or other rational segmentations of the market.
• The company is developing products and services that leverage expertise and rely on a subscription revenue model. (They are not simply hosting their application.)
• The company has a persuasive way of measuring the value its products and services generate and of demonstrating that one can get value reliably by engaging with them.
• The company differentiates itself from the Borg by being more open and honest about the difficulties in gaining value.
• The company has abandoned the traditional rhetoric of the crossing-the-chasm model, recognizable in all press releases by the phrase, "the leading provider of XYZ software."

In coming weeks, we'll look at a number of best-in-breed companies and see how these broad indicators apply in each of their cases.

We will also see, I believe, two things that I have only touched on here. First, the growth pattern for a company that is adopting a new-model best-in-breed strategy is quite different from that of the company that is tyring to cross the chasm. For the most part, we will be looking at the relatively mature companies that have established customer bases and a steady stream of maintenance revenue. But it will be worth while to touch on what's happening at earlier-stage companies.

Second, we will talk a little bit about SOL, oops, I mean SOA. For these best-in-breed companies, a new generation of the Borg based on services-oriented architecture is a two-edged sword. On the one hand, it attacks them in their weakest point, their platform. But on the other hand, it could also make it easier for them to coexist.

To see other recent Short Takes, click here for a listing.

Why best in breed can work.