ArsDigita Archives

Supporting an Open-Source Software Products Company with Service Revenues

by Allen Shaheen (, Philip Greenspun (

Submitted on: 2000-01-01
Last updated: 2000-09-01

ArsDigita : ArsDigita Systems Journal : One article

Allen Shaheen was a co-founder of Cambridge Technology Partners. In 13 years at Cambridge, Shaheen repositioned the business from training to systems integration, expanded globally, and participated in its successful IPO. At the time of his departure for ArsDigita, he was Executive Vice President of Cambridge's international operations: 1,200 employees in 17 countries with $145 million in revenue. Shaheen is CEO of ArsDigita.

Philip Greenspun is a computer science researcher at the Massachusetts Institute of Technology, specializing in software engineering for Web applications. He founded ArsDigita in 1997. His personal Web site is

In the year 2010, all enterprise software will be free and open-source. The information system needs of organizations are diverging and packaged solutions are becoming increasingly less viable. The closed-source packaged software products that generated so many billions in profit during the 1970s and 1980s were dependent to a large degree on stable business processes and information system needs. Pressure from the Web revolution pushed IT project schedules from 3 years down to 3 months and closed-source packaged software, with the exception of commoditized standards such as relational database management systems and operating systems, has thus far played an insignificant role in building the new economy.

In a world of diverging information system requirements, open source is an unstoppable competitive edge. A closed-source system can only solve those problems that its designers envisioned. When new business processes are developed semi-annually, it is impossible for software that was designed six months earlier to address the new needs. An open-source system won't meet the new business needs out of the box, but it can be rapidly extended for the adopting organization until it works for their specific needs. As information system designs continue to diverge, organizations will become increasingly skeptical about the claims of any software vendor. Instead of vendors getting paid for shipping a CD-ROM that customers start using, enterprise software companies will be paying customers to invest in experimenting with their proposed solution.

Where software has become standardized and commoditized, closed source systems are much more effective solutions. However, as hardware becomes ever cheaper the transaction and actual costs of licensing closed-source systems become glaring. In the presence of a well-specified standard, such as the Unix operating system or the SQL language, the revolutionary coordinated distributed programmers method that produced the Linux operating system are extremely effective and eventually will produce an acceptable competitor to closed-source packaged systems. An acceptable, free, and infinitely flexible open-source solution will be an irresistably attractive competitor to a traditional licensed software vendor.

We started ArsDigita in the late 1990s, towards the end of a glorious 30-year period in which fabulous wealth was generated from closed-source software products. Conventional wisdom, freely and forcefully offerred to us by a range of industry pundits and venture capitalists, would have been to sell packaged software and hang on to the source code. Our collaborative commerce system, the ArsDigita Community System, would then join the ranks of database-backed systems supporting the Web sites of a handful of Global 2000 companies. As the market shifted toward open source, we'd adapt to making a profit in that market.

We rejected this wisdom. Our goal was to build a strong standalone enterprise software company for the Year 2010. We wanted to get very good at inhabiting our adult form and not spend time and effort being successful at a larval stage with closed-source license revenue. Eschewing the easy path necessarily put pressure on us to develop a strong and growing service revenue stream. After all, if the software is free, just how can one make money? This article explains how revolutionary service revenue growth was achieved at Cambridge Technology Partners and how we are doing it at ArsDigita.

The Early 90s

In 1991, a group of senior executives at Cambridge Technology Partners (CTP) was figuring out how to break past the traditional model of the professional services business. At the time, CTP had the following strategic strengths:

  • Fixed time/fixed price business model
  • Unique rapid application development methodology
  • Focused expertise in scarce client server technology

The problem we faced was growth and scalability. We felt that the long-term survival of the company depended on reaching and sustaining a 50-100% compound annual growth rate, in an industry where firms were typically growing at half that rate. Growth by acquisitions was a possibility, but we felt the impact on company culture would be too severe.

In the end, we decided that CTP could achieve accelerated growth by two parallel means: specialization and a product-distribution methodology.

Specialization requires identifying the key processes that define an IT service company and creating specialized groups within the company to excel in the execution of each of them. The glue that ties these specialized groups together is a culture of high quality service for clients, satisfaction for employees, and aggressive growth for shareholders or partners.

The product-distribution methodology -- essentially, adapting the strategies of a product company to a service business -- requires four things: a clear message of what solution the firm is world-class in, a distinct marketing organization for building market share and mind share, a distinct sales organization, and a "productized" front-end service offering that is easy to articulate and sell to clients and simple and low risk for the sales organization to sell.

The rapid growth and scaling enabled by specialization and a product-distribution methodology has meant that IT services firms can achieve critical mass in a relatively short period of time. In five years CTP grew from 120 people and $12M in annual revenues to over 2000 professionals and $300M in revenues. And over the past 10 years, the number of professional IT services firms publicly traded or targeting IPOs has increased from a handful in 1990 to over 40 in 2000. The ability of these firms to gain critical mass rapidly with both clients and staff using the methods described in this article has been critical to their success.

Conventional Thinking

Historically, professional services organizations have been partnerships. In this structure, the organization was built upon the foundation of highly skilled individuals groomed and promoted on the basis of their expertise developed over a long apprenticeship. Senior partners nurture and develop client relationships over a long period of time. By virtue of their long service, these partners have deep knowledge of their clients' business and strong relationships within the client organization. The model requires that partners run the business. It also requires that each partner is an "all-around athlete" skilled in operations, business development, general management, and client relationship building.

Growing a professional services organization via the partner model is a structured process with a clear direction and path. For the employee, the goal is to become a full partner charting the direction of the organization and sharing in its profits. An employee must build experience in the operations of the firm, gain delivery management experience, build up solid client industry expertise, and round out that experience over time with business development skills. Since the process is structured and lengthy, the growth rate of the firm mirrors the lengthy apprenticeship process.

In such firms the growth of the organization as well as its economic model is often described by the pyramid structure, which models both the ratio between junior and senior firm members (partners), and how these different levels of staff are deployed on client projects. The pyramid also represents the number of client relationships, project teams, and general staff that a partner manages as part of his or her practice.

How many individuals does the pyramid encompass? In many of the larger IT professional services firms, a partner is typically expected to "leverage" a group of between 40 to 80 professionals. Thus, the pyramid has one partner at the top developing business, managing a series of client relationships, and managing a staff that in turn manages and delivers engagements.

Also displayed are the core processes on which each staff group focuses:

  • Partners typically focus their efforts on business development and client relationship activities.
  • Managers focus on client expectation management and successful project delivery.
  • General staff are deployed to deliver projects as directed by the management staff.

The New Paradigm

In the early 1990s a number of professional services firms led by CTP challenged their industry's idea of sustainable growth rates. These systems integration and software consulting firms saw that the growth rate of a traditional professional services firm was constrained by its ability to hire, develop, and retain partners -- a difficult and lengthy activity, because partners were expected to be world class in all the core processes of delivery, project/program management, and business development.

These relatively new professional services firms aimed to scale rapidly by building separate world-class groups around each core process (delivery, management, and business development) and applying distribution strategies long used by product companies.

Splitting the Pyramid

The first strategy, specialization, replaced the traditional pyramid of the IT services firms with three interrelated pyramids, representing separate groups within the firm as shown in the figure above. Each group (business development, management, and delivery) would aspire to achieve excellence in their respective area and grow their area as rapidly as possible by focusing on their core activity.

Applying Product Distribution Methodologies

CTP and other young IT services firms (such as Sapient Corporation) also applied the approach of having a business development group distinct from operations. At CTP, senior management soon realized that the success of the business development group would be greatly enhanced by applying distribution methods long used by product companies:

  1. A clear message of what solution the firm is world-class in
  2. A distinct marketing organization for building market share and mind share
  3. A "productized" front-end service offering that was easy to articulate and sell to clients, as well as being simple and low risk for the sales organization to sell
  4. A distinct sales organization

Let's examine each of these methods in our industry context.

Building the Message

In order to succeed the firm needs a message that is clear and easily understandable by four groups:

  • Clients and potential clients
  • Employees and recruits
  • Shareholders
  • Industry analysts and press

The message must clearly and concisely state what the firm is world class at -- a statement that differentiates this firm and its methods, people, and tools from the crowd. CTP, for example, articulated its focus as "Customized Rapid Development of Strategic Application." In many cases we shortened this to "Rapid Application Development," a crisp statement of the firm's expertise. We defined "Strategic Applications" as revenue-creating applications rather than those intended purely for cost-cutting. Examples: executive decision support systems, customer relationship management applications, order entry, order management, and sales force systems.

In addition to a clear message, the firm needs to articulate its perspective on the market, which includes the historical context, need, and benefits associated with its chosen market focus. For example, ArsDigita Corporation is a rapidly growing enterprise software firm that builds software solutions to enable "Collaborative Commerce." The historical context comes from the development of the Web as an enabler of on-line communities, and the power associated with rapidly bringing people together to share experiences, opinions, and general content via the Web. The need comes from the fact that users revisit sites that have a purposeful community associated with them, where they can interact with others of similar interests, and where the site is valuable purely for its information and content. The benefit comes from increased site traffic, longer interaction time by users, greater revenue generated, and ultimately a higher degree of user loyalty to the site based on the depth of the community and the richness of the content.

Broadcasting the Message with World-Class Marketing

Once a firm develops message, differentiation, and perspective, it must distribute them to a wide and broad audience. The goal is to build up mind share so that the perception of the firm is at least 10 times larger than the actual market share, creating future demand into which the firm can grow. The mass distribution of the message is the responsibility of a distinct marketing organization. Marketing is also responsible for gathering and distributing to the sales organization the benefit of mass distribution, namely, business leads and potential new clients of the firm.

CTP achieved mass distribution of its message through:

  1. Highly effective partnering programs
  2. A massive series of solutions-focused seminars
  3. An aggressive public relations campaign focused on industry analysts and the press

CTP developed its most effective partnering programs with hardware and infrastructure companies such as Sun and HP and with application software vendors such as Seibel Systems. (Attempts at partnering with management consulting firms were generally unsuccessful.) Such programs were very focused on business development and were measured on the number of clients or potential clients that were touched by the process, the number of hard leads developed, the number of closed deals, and ultimately the revenue value of those deals. Depending upon the scale and importance of the relationship, one marketing person was typically responsible for 2-3 major relationships.

Next, CTP focused on the development and delivery of sales seminars, focused on a particular need or opportunity such as "improving customer service responsiveness" or "accelerating product or program launches." The seminars were jointly marketed and delivered over several months in dozens of cities with the local operations of both the firm and the partner organization supporting with speakers and cases, and corporate content and coordination provided by the marketing group.

The industry analyst and press relations were also handled with dedicated marketing staff, with help from external PR agencies. This internal PR group was dedicated to developing 3-5 analyst relationships deeply and ensuring that press releases were coordinated and timed with major events. CTP, for example, chose quality over quantity, developing strong relationships with a few industry analysts able to understand the company, its strategy, and its services well.

Finally, marketing was tasked with sending the press an almost constant stream of news items, drawing from client contracts, client launches, new office openings, participation on industry panels, internal promotions, strategic hires, and the achievement of general growth milestones.

A Service that Looks Like a Product

Building an efficient distribution channel with both industry partners and an aggressive sales force requires that the firm's service offering be easily articulated, focused, and priced. Professional service firms like CTP had a problem in that their fundamental service was custom software development. The fact that every application was made to meet the individual needs of a particular client demanded a more complete understanding of the clients' business, and a more complex articulation of the service.

Take the bidding process as an example. At CTP, bidding a project accurately required specialized experts within the company who were highly leveraged and unavailable for every sales and partner call. CTP solved this problem by developing two well-defined and short duration service offerings. These services lasted 1 to 3 weeks and resulted in a clear set of deliverables that were easy to articulate and did not vary greatly from client to client. These services were referred to as the "Scoping" and "Rapid Solutions Workshop" (prototyping) phases of the project, used early on in the client relationship at low cost. The aims were to:

  • Deliver value to the client commensurate with the fee
  • Qualify the client as having a realistic budget for the project
  • Allow experts in project sizing to get a good view of the project scope and develop credible time and cost estimates
  • Further the relationship with the client
  • Deepen the knowledge of the client's business issues

These two services enabled the sales organization and partners to sell CTP services rapidly and easily without requiring extensive support from scarce operations people. The risk of selling a poor piece of work was reduced by the fact that these services were of a short duration, had clearly defined deliverables, and could draw on extra professionals as needed for the short term.

Thus with a clear set of deliverables derived from the twin exploratory services, the sales force could be confident in what they were selling, the operations group could be comfortable that they understood the scope of work needed, and management could be comfortable that there was a low risk of a major financial problem due to a poorly sold or managed engagement.

Taking it to the Market

Ultimately, building a rapidly growing IT practice requires an aggressive and focused sales organization. All of the previous activities are meant to make the sales organization's job easier, faster, and more efficient.

In the traditional world of partner-driven organizations, client acquisition was left to the partners of the firm. The new model required that an effective sales organization be built up rapidly to fulfill the role partners would have played. At CTP, we developed a geographically based, professional sales force in a very short period of time with an aggressive hiring campaign. We offered sales professionals lucrative commission plans, stock options, the opportunity of selling strategic solutions to clients, and the chance to grow their own geographic sales region. This was an innovative concept at the time and enabled CTP to attract a large number of sales professionals quickly.

CTP's standard plan for the development of a new office called for a local sales person to be hired first, typically a seasoned sales professional with knowledge of the IT industry, extensive local (geographic) contacts, and a proven sales track record. This strategy enabled CTP's rapid growth, as there were always new geographic markets to enter.

Initially, the sales organization was focused towards the development of new client relationships rather than expanding existing client relationships. Once the first major project with the client was sold, the sales person was driven by the compensation plan to seek new business with a new department of that client, or a new client altogether, rather than continue to work on enhancing the current relationship. Although this program was highly effective in the early stages of CTP's growth, it was not conducive to long term relationships with clients or to the long term growth of the company. Better programs organize their business development functions around industries or specific application areas of interest to clients. Examples of application areas are: customer relationship management, e-commerce, or on-line community development.

Structuring the sales force along industry lines allows it to develop more expertise and relationship knowledge directly related to its target client base. However, this strategy requires more investment in personnel and training as industry expertise is generally scarce, comes as a result of experience and extensive training, and ultimately requires a higher compensation structure than hiring and retaining generalists.

Shortcomings of an IT Service Business

We've painted a rosy picture of an IT service business. Now for the warts. A pure services company spends most of its time reinventing the wheel. The San Francisco office doesn't know that the New York office has already solved a similar problem so it starts from scratch. The Singapore office is working for a customer that uses Brand X middleware and the Paris office is working for a customer that uses Brand Y middleware. They each learn the quirks of their respective tools. A multi-office IT service business may be more a loose federation of programmers than a community of practice. Worst of all, revenues only scale by adding bodies.

Why a Product + Service >> Pure Service

A company that combines an open-source product is much healthier than a pure services company. First of all, the open-source product functions as a knowledge management system. The best knowledge management systems involve several layers of refinement, approval, and ultimately publication. This happens to be exactly the process by which lines of code turn into a released software product. Generalizing, productizing and releasing code that had its genesis in the course of a services engagement is an extremely effective way of communicating the reusable knowledge that was developed during that engagement.

A second benefit of services centered around a product is that it tends to push all the people in the organization toward the same set of tools. At ArsDigita, for example, we have deep expertise on three continents in how to achieve 24x7 reliability of Web services based on Solaris, Oracle, and the ArsDigita Community System. Quite a few problems have stumped project teams but none have proved beyond the reach of the project team plus help from an expert in another ArsDigita office. How did the project team find the right expert within our 230-person organization? The expert's name was on a piece of source code in our product or on the documentation for a module.

Finally a product+service company has unique opportunities to improve productivity and therefore revenue per programmer. Every time we add new features to the ArsDigita Community System we are adding value for the customer. A pure services firm can add value by training people or giving them more experience but those efforts may prove illusory in a world of highly mobile workers. The value and therefore prices of a community of practice should go up as the quality of the tools and methods used by that community is improved.

At this point the thoughtful reader will ask "Why do you have to release this product externally? Could you not achieve all the same benefits by having an internally developed and internally released code base?" Theoretically it would be possible to build a community of practice around an internal-use only product. However, consider how such a product might evolve shielded from the discipline of the marketplace. Also, why would a really strong software developer want to work on a product that hardly anyone ever got to use? Programmers have big egos and want the world to appreciate their accomplishment. Note that Microsoft has no trouble attracting great developers. They are confident that their work on a Microsoft product will be widely distributed. An open-source product is even more attractive to programmers. They can be sure that their work will get the widest possible distribution. This is such a powerful lure that thousands of superb programmers worldwide have voluntarily contributed to Linux.


IT service revenues can scale rapidly, as has been demonstrated by a number of such pure IT service firms that are now publicly traded. The main strategy behind the rapid scaling of IT service revenue has been a fundamental reconstruction of the traditional professional services partnership structure, such that the three core functions of delivery, management, and business development are separated into distinct, specialized organizations within the firm. At the same time, adapting and deploying product-distribution methodologies can very quickly expand the market and client base of service firms.

ArsDigita's early growth from $2 million/year to $32 million/year in revenue in 18 months shows the inherent natural power of service revenue growth tied to an open-source enterprise software product.

The most powerful software firms of the coming decade will be ones that combine a universally adopted open-source enterprise software product with the best practices learned from the pure IT services firms.

Reader's Comments

I am curious what feedback or incentives are present (or were present) to help ensure that the estimates, requirements, schedules and costs developed in the "Scoping" and "Rapid Solutions Workshop" accurately reflected the actual tasks and costs of the project upon Launch.

I worry that if performed as part of an initial sales phase, by groups that are incented for sales and not responsible for the outcome of the project, there would be incentives to make actual costs seem lower than they are in order to "win sales."

If you want delivery of the actual service to be profitable for the company, and well received by your customer, than it is necessary for these scoping tasks to be performed by a team that receives "negative feedback" (in the control theory sense), or incentives for accuracy (in the IBM sense).

This would seem to make the splitting of these pyramids into separate teams of specialists more difficult.

This is a problem that many organizations share. Can someone comment on such issues have been addressed?

-- Jerry Asher, October 24, 2000

I believe that contrary to your argument your firm should pursue a two stage ("larval v. adult") development cycle culminating in a closed-source product as follows.

Larval Stage

You face these problems as a small firm

1. need to differentiate in a very crowded market
2. need to reassure customers that the fledgling product will be supported
3. need to motivate laborers to build this product

Therefore you should embrace open source

1. because an open source enterprise solution is different
2. because customers may be assured that they can support the product if your firm fails
3. because you may harness the idealism of your laborers to stimulate their productivity

Adult Stage

You face these problems as a big firm

1. need to seek a broader market
2. need to increase profit margins
3. additional demand for labor
4. need to control quality

These problems will be better addressed with a closed-source version of the software

1. because you can license it for to third-party distribution channels
2. because the profit margin on licensed software is higher than that on service labor
3. because the internal culture of the firm will stimulate productivity
4. because you can better control software quality than service quality

Managing the Transition

The transition from larval to adult stage could cause a loss of identity as sure as that faced by a caterpillar which becomes a moth.  This loss of identity could be significant if it disrupts the firm's differentiation in the marketplace or reduces its employee morale.  This problem may be addressed by transitioning the codebase to a different set of technologies supporting binary compilation while leaving behind a basic functional open source shell.  This shell may function as the caterpillar's coccoon and give comfort to those customers and employees who miss the caterpillar itself.  Meantime marketing must focus on differentiation the newly minted moth.


It must be noted that this plan only works if your product is not dependent on a classic open source model.  Linux, Apache, Perl, and other classic open-source products were notable because a distributed group of self-managed laborers successfully tackled broad technical problems without any obvious financial incentive.  These products are dependent on a grass roots development community.  By contrast, your firm employs a cloistered group of well-paid and well-managed developers (see "Managing Software Engineers" by Philip Greenspun in the ArsDigita Systems Journal, to serve a small but profitable industry niche.  Therefore your product is not dependent on the open source developer community.  I believe this ensures that your firm may transition from an open source to a closed source product without substantial harm to your enterprise.  Your closing observation that the benefits of open source are theoretically acheivable with a closed source product borrowing open source methodology would seem to indicate that you support this conclusion.  In any case, the discipline and exposure of the market will hew either type of product equally.

-- Alec Permison, November 8, 2000

I have to disagree with the last comment about going closed source. I worked at Vignette which is a closed source competitor and the closed source development model has some serious drawbacks. The main one is that the internal engineering staff developing code is isolated from field staff implementing the software solutions so good lessons learned in the field are not easily incorporated back into the product. There is also cultural isolation with field consultants not being able to participate easily on internal projects. There is also the tendency to take any additions developed and charge for them which often irritated clients who had just shelled out half a million dollars to find they had to pay more for even small features. Since software tends to be commoditized over time, the license revenue can only be maintained by increasing the complexity of the software which decreases reliability and customer confusion. While Microsoft has succeeded in this , it unlikely to be a successful strategy for most companies As far as ACS being solely developed internally, that is just not the case. There is a sizable developer community which uses ACS either in the Classic or OpenACS versions and one reason is the full availability of the code. Closed source products have much more limited developer communities as no one can afford the software and trial licenses are generally so limited that they arent worth the time.

-- James Ross, February 19, 2001