4. 3. 1  Specialist/vertical (a free application as the main product)

In this section, we include companies that produce free software as the promoters and/or leaders of specific projects. Their involvement with free software is thus very significant and one of the key aspects of their business strategy will be the management of the community and seizing of the opportunities for innovation, diffusion and volunteer work that it offers. In essence, these models have a free product for the community and a product or related service as their commercial offer, so the key to their success is often to strike a balance between the two. According to Marten Mickos, CEO of MySQL AB:

FOSS companies will not work unless they serve equally those who want to spend time in order to save money, and those who want to spend money in order to save time".

These companies are the most common in Daffara's study, including the first four categories (twin licensing, OSS/proprietary versions, badgeware and product specialists). They equate to the product companies we saw in module 3 of this subject, so their main problem will be how to recover the initial investment in development.

As we saw in the above classifications, a common strategy is to obtain revenue through proprietary licences, which are combined with free licences in different ways.

There are also models that cyclically combine proprietary licences, like Hecker's "loss leaders" and "sell it, free it". The loss-leader concept is not unique to software, being a widespread strategy in every sector of activity: a product is offered free of charge – or at such a low price that it entails losses for the supplier – as a way of attracting the attention of a large number of potential customers to whom the company intends to sell other items. Hence, both dual licence models and free products with proprietary extensions use a loss-leader strategy to some extent.

Besides promoting the sale of the related product, there are several benefits to adopting an open-source strategy of this nature in the software industry, such as helping to establish the technology as a de facto standard, attracting improvements and complements to make the product more appealing, generating sympathy in an audience that includes potential customers of the related product, and reducing the maintenance costs of the project.

We will now look in detail at the twin- or dual-licensing model and the model combining a core free product with proprietary accessories. We omit other models in which the main product is not free because they are really business models based on proprietary software: the code is only released as a complementary business strategy to enhance the position of the core proprietary product.

Daffara also lists several companies that carry out development projects with entirely free licences and earn their income from ITCS(installation/training/support/consulting). This group is perhaps one of those that can encompass the most different models, since its revenue source is a rather vague category. Hence, it is important to look closely at the markets they serve and their differentiation with equivalent products, in addition to best knowledge.

4. 3. 1. 1  Mixed models: dual licensing

This model is based on the distribution of a product under two different licences: a traditional proprietary licence and a restrictive free licence (GPL type). Thus, if somebody wishes to derive a work from it and redistribute the new work without the code, they can, but they must pay for a licence. Otherwise, all derivative works must be redistributed with the code.

Michael Olsen, manager of Sleepycat Software Inc., producers of BerkeleyDB, describes its dual-licensing model thus:

"The Sleepycat open source license allows the use Berkeley DB [...] without cost, under the condition that if the software is used in an application that is later redistributed, the complete code of the application must be available, and must be able to be redistributed again freely under reasonable conditions. If you do not want to offer the source code of an application derived, you can buy a Sleepycat Software license."

S. Comino; F. M. Manetti. "Dual licensing in open source markets". Available at: http://opensource.mit.edu/papers/dual_lic.pdf

This strategy is appropriate when a substantial proportion of the demand is generated by commercial users who need to embed the software in their own products. These customers use the product purchased as input for the production of new software, either as an end product or as part of a more complex technology produced and sold by the commercial customer. Whether because they need to be able to sell their derived products under a traditional proprietary system or because the software they generate is a fundamental part of their differentiation, this customer will need to close the code it generates and must therefore pay to do so.

These models divide their users into two groups: the community – all users who are content with free licences, and use the product under these terms – and corporate clients sensitive to the reciprocal terms of free licences.

However, maintaining a community of people collaborating on the product can be problematic. On the one hand, if direct income is obtained through the product, this could affect the motivation of the volunteers who contribute without receiving anything in return. While on the other hand, the companies that implement it must formally obtain copyright assignment from the volunteers in order to avoid future problems from disgruntled employees claiming their share of revenue from licences for the product that they helped develop.

In practice, companies that base their model on dual licensing do not benefit greatly from the possibilities of external contributions to the development, obtaining only small-scale debugging and the odd patch from the community. The main development team is typically almost 100% dominated by company employees.

Another problem that can arise with these models is that their customers can build their own proprietary extensions without modifying the original code, so they can use the free licence version and have their add-ons as a separate and independent application.

These companies often combine the revenue generated from dual licensing with other activities such as the provision of services, which we will see later. Examples of this model include Funambol, MySQL, Sleepycat DB, and TrollTech/NOKIA.

Corporate data on Funambol, Inc. Table prepared with statistics from Hoovers (http://www.hoovers.com)

Company name

Funambol, Inc.

Head office

Redwood City (United States)

Website

www.funambol.com

Creation date

2001

No. of people employed in 2007

40

Turnover in 2007 (million)

$4.8

Funambolis a US corporation engaged, as its motto states, in "mobile 2.0 messaging powered by open source". The company develops a mobile application server (providing push e-mail, address book and calendar, data synchronisation, and an applications server for mobile devices and PCs), together with a development platform for mobile applications, both developed under the name "Funambol".

It sells its code base under two licences: the AGPLv3 for its "Community Edition" and a commercial proprietary licence for its "Carrier Edition". It also combines this strategy with providing the additional functionality required for large-scale implementations of the closed version as well as services based on the "Carrier Edition.

In this case, Funambol chose the "Affero" GPL licence, which affords it extra protection against the commercial use of its applications in the form of software as a service (SaaS). As discussed in module three of this subject, the GPL allows the code to be modified without redistribution, provided that the application itself is not redistributed, as occurs with the provision of software as a service. The "Affero" GPL solves the problem of this void by requiring redistribution of the source code when the software functionality is offered under the "SaaS" model too.

The nature of the software makes it an ideal candidate for the dual-licensing model as it appeals to other companies seeking to develop closed applications on its platform, such as mobile telephony operators, device manufacturers and other software companies. Because of its use of the AGPL, companies that use Funambol as the basis of their "SaaS" offers must also pay if they do not wish to redistribute the code. Its customers include Vodafone, Earthlink and Computer Associates.

Funambol tries to fully exploit the needs of large corporate customers with the incorporation of additional functionality in its commercial version and services. To avoid the problems of the "free kernel + proprietary accessories" model that we shall see later, it makes sure that the closed functionality is only interesting in scenarios of large corporate implementations, so its free user community will not feel the need to develop this functionality by itself.

In "My Honest Dual Licensing", Fabrizio Capobianco, manager of Funambol, argues that the dual-licensing model is the most "honest" model in upholding the principles of free software development, making it more compatible with business' need to generate profits.

However, as we saw earlier, defining a viable source of income does not guarantee the success of any company and the use of free software will allow us to implement qualitatively different strategies to those of a model based on proprietary software. Funambol is a case in point here, since the company had to refine its marketing practices and their target populations before hitting on a viable business model.

In its early days, Funambol tried to set up a classic software-vendor model around its free software product. The company developed Sync4j, which enabled developers to build applications for mobile devices under the "sometimes-connected" paradigm (the application can work offline, synchronising data when the connection is restored). It identified large companies and wireless operators as prospective clients which, due to large staff numbers and the increasing opportunities for mobility, would need to synchronise data between mobile devices and their corporate servers.

In order to reach these customers with its product, Funambol decided to pursue a proactive sales strategy with particular emphasis on marketing and its sales force, which tried to access the potential customers directly through telephone campaigns.

It met with very limited success. Funambol failed to meet its sales expectations because it found that large corporations were reluctant to deal with small new companies. In addition, the sales cycles were very drawn out and it soon became apparent that a much bigger sales and marketing team was required to maintain this strategy than Funambol could afford.

Funambol quickly realised that its problems were down to this active sales strategy, the traditional method in the world of proprietary software but a barrier to entry that only a handful manage to overcome: to access a group of potential customers consisting of large corporations, it is often necessary to have a large sales and marketing capacity, as well as a sufficient size and reputation to deliver the required confidence.

The use of free software allowed the company to reverse this strategy, focusing on a reactive type of marketing in response to initiative from the customer. In this new scenario, potential customers would seek out Funambol, leaving the company with the role of being attentive in order to identify them after contact.

The effectiveness of this strategy depended on a single factor: the number of downloads of its product. Once it had obtained enough downloads, it was able to identify the following typical sales cycle (much shorter than the one observed with the previous strategy):

  1. The potential user accesses the Sync4j website for information on the product and technical documentation.

  2. The user downloads the product.

  3. He/she later subscribes to the mailing list for more information.

  4. Following extensive use of the product (usually in R&D projects), the customer contacts Funambol to ask about pricing and licence conditions. Internally, they are classified as prospective customers.

  5. Finally, they ask for a quotation and formal offer and can become a Funambol client.

(Fabrizio Capobianco; Alberto Onetti. "Open Source and Business Model Innovation. The Funambol case". Available at: http://oss2005.case.unibz.it/Papers/4.pdf)

The key factor for continuing to fuel this cycle is, as we explained earlier, maintaining a high number of product downloads. The cycle is continuous, generating more downloads by itself. Thus, after the initial effort, this mechanism gathers enough inertia to work virtually alone.

To do so, Funambol concentrated on creating a community around its product, focusing its marketing efforts on the users of its free version, both experts and those with fewer technical skills. Although this strategy is not directly oriented towards its revenue-generating customers, it proved much cheaper and more efficient.

The company focused on raising the profile of the product among developers, participating in development forums, mailing lists, specialist publications, conferences, creating partnerships with non-profit organisations that promote free software and establishing synergies with other well-established open-source products. For more inexperienced users, it had to ensure that the product was easy to install and that sufficient documentation was available on the website. When it began to focus on these last two factors, the company observed a substantial increase in the number of product downloads, thus setting in motion its sales-generating cycle.

Recommended website

For more information, see:

http://www.funambol.com/blog/capo/2006/07/my-honest-dual-licensing.html

4. 3. 1. 2  Mixed models: free product kernel and proprietary accessories

In this model (Daffara's "Split OSS/commercial releases"), a program has two different versions: a free basic version and a proprietary commercial version based on the former but with additional functionality implemented through plug-ins or accessories. The free version must use an MPL or BSD type licence allowing the combination in order to create a closed product.

The main problem with this model lies in keeping the free product interesting enough without taking value away from the revenue-generating proprietary product. We also run the risk that the community formed around the product may decide to develop the functionality of the proprietary version on its own, making it difficult to generate revenue from sales.

In this model, we can distinguish between two classes of users: those who are willing to pay for a product with some additional features (medium and large companies), and those who are very sensitive to price, such as small businesses, micro-enterprises and private users. By combining free and proprietary versions, we obtain a more widespread adoption of the proposed solution without missing out on revenue capture from proprietary versions. As we saw in previous modules, in a "winner takes it all" scenario, common in software, strategies based on widespread adoption are very important.

Hence, it is based on the same user-segmentation principles as the dual-licensing model but is more at risk of losing the sympathy of the community, since it does not have access to the entire source code.

An example of this model is Sendmail Inc., which sells an array of proprietary products around the sendmail open server. Other examples include Hyperic (IT Operations/Monitoring), SourceFire (SNORT commercial version), Zimbra/Yahoo (messaging, groupware) and XenSource/Citrix (virtualisation).

Corporate data on Sendmail, Inc. Prepared with statistics from Hoovers (http://www.hoovers.com)

Company name

Sendmail, Inc.

Head office

Emeryville, CA.

(United States)

Website

www.sendmail.com

Creation date

1997

No. of people employed in 2007

125

Turnover in 2007 (million)

$23

When studying business models based on free software, we often think of corporations that decide to open up their code as a competitive advantage to expand their market share. Sendmail is an interesting case as this process occurs in reverse: with free, non-profit roots, the creation of a commercial initiative around the project is aimed not only at generating revenue from the development, but also to maintain the project's dominant position in its sector and to expand its user base.

Sendmail is a mail transfer agent (MTA) and one of the best known examples of projects born out of free software communities. In 1998, it was estimated that 80% of all e-mail traffic was sent through Sendmail. It is still the most popular MTA on the Internet, although it has lost some users to Microsoft Exchange Server, Exim and Postfix. Equally important is the long lifespan of the product, whose origins date back to developments started in the 1970s.

Eric Allman developed the first version of Sendmail at Berkeley University in the early 1980s on the basis of previous work on the Delivermail program and founded Sendmail, Inc. in 1997. The company strategy focused on selling additional Sendmail functionality in a proprietary format (e.g. user-friendly interfaces) in addition to providing complementary services. At the same time, the company made an effort to openly maintain the continuity of Sendmail's development by providing hosting services and human resources for its development.

When he set up the company, Allman expected not only to develop a business, but also to protect Sendmail's dominant position, which was being threatened by the emergence of proprietary formats that jeopardised the SMTP open standard. The company concentrated its efforts on the corporate environment, offering not only integration and support services, but also a product that was more responsive to its needs. The extensions created by the company provide graphical interfaces and ease of management, and are marketed in proprietary formats.

"Sendmail, Inc. develops commercial products and services for ISPs and enterprises for whom email is mission critical, while continuing to drive innovation and standards through Open Source software development."

Sendmail, Inc

We can consider the creation of Sendmail, Inc. to have been the necessary step to cross the "chasm" and guarantee the product's adoption by the pragmatic and conservative majorities. Nonetheless, for Allman, it was important to maintain the original functionality of free Sendmail, so Sendmail Consortium was set up as a non-profit entity to develop the free version. In this way, it can capitalise on the advantages of an open development model, such as contributions, cost-cutting, product innovation and evolution.

Allman thus took advantage of "the chasm" to sell proprietary extensions to his product without the danger of forking his project. Following Moore's model, the community around the free Sendmail project consists of innovators and technology enthusiasts interested in the raw functionality and new proposals. Business customers, however, are pragmatists and conservatives with very different needs and aims. The proprietary extensions, which focus on the functionality of the product packaging and finish (ease of use, graphic interfaces, stability, etc)., are not only uninteresting to innovators, they may even seem unnecessary. The presence of this chasm between the interests of the community and commercial customers allows for the co-existence of the core free version and the widespread proprietary version without the risk of forks, since the community has no interest in the extensions on the other side of the chasm.

4. 3. 1. 3  Free models: "distributed sale" of the product

It is commonly assumed that licensing a product in free format leads to loss of opportunity for earning direct revenue from the intellectual property rights over it, creating the need to exploit complementary products or services.

However, choosing a free licence for a project does not necessarily mean forgoing the possibility of obtaining revenue directly from this product. The widespread idea that nobody will pay for something they can obtain for free does not paint a true picture of reality. Many people are willing to pay a small sum for a work that they value if they think that this money will go to the original authors. If a project is successful enough, it may receive small contributions from a lot of people, perhaps even managing to fund its creation in the same way that a street artist does not charge admission but can raise enough to make his or her investment in time and effort worthwhile. This is the idea behind theThe Street Performer Protocol and Digital Copyrights, by John Kelsey and Bruce Schneier, which proposes a distributed funding mechanism for digital works in which the author does not complete his/her work until sufficient funding has been collected.

Recommended website

http://firstmonday.org/htbin/cgiwrap/bin/ojs/index.php/fm/article/view/673/583

Different mechanisms have been described and implemented for structuring this direct, distributed funding in the context of software development, from grants and bounties to the creation of on-line markets that bring together developers and prospective clients, based on a bonus scheme similar to that described by Chris Rasch inThe Wall Street Performer Protocol.

Recommended website

http://www.firstmonday.org/issues/issue6_6/rasch/index.html

Donations are the most straightforward mechanism for this type of financing, but too unstable for creators, who need the security of an income before they invest their time. In bonus and bounty systems, the people interested in a specific functionality offer a reward for it to be implemented. When the total reward – which various people can contribute to – reaches a sufficient sum for a developer, he or she can offer to do it and is paid once it is finished. Some of these systems rely on the trust between the development team and the users, and have no payment guarantees, while others propose the establishment of some form of neutral intermediary.

The Cherokee server

This server decided to implement a bounty system with the primary aim of attracting new developers to the project. Besides rewarding effort, providing a financial incentive would attract more people to the development community and encourage the growth of the project.

Virtual markets

Several attempts have been made to create "virtual software markets" based on this type of funding. Some of those currently operating include BountyCounty (http://bountycounty.org/) MicroPledge (http://micropledge.com/) and BountySource (https://www.bountysource.com/).

The key to success in these scenarios may lie more with the payment facilities offered than with the willingness of users to pay:

"Most people are happy to pay a tiny extra bit on top of some larger amount, if they have their wallet out already and think it's for good reason. When people fail to make small, voluntary donations to a cause they like, it's more often due to the inconvenience (writing a check, putting it in the mail, etc), than the money.

(Karl Fogel. "The Promise of a Post-Copyright World". Available at: http://www.questioncopyright.org/promise)

Although many projects implement these ideas to obtain additional funding, it is difficult to identify corporate scenarios where the bulk of the revenue is obtained through these mechanisms.

Firstly, in the context of software, this type of funding can be more difficult to obtain because of the absence of a strong identification with and sympathy for the authors, which does exist with other creative works.

Secondly, this model is likely to be more successful if it is a non-profit free software project composed entirely of volunteers, which will arouse the sympathies of its users more easily. A company wishing to use it successfully will no doubt have to obtain prior acknowledgement through transparency and trust, proving that profit-making is not the be all and end all and that the project will have an impact on the common good (we will look later at business models based on these principles).

These systems have a more direct economic model, eliminate intermediaries and ensure greater proximity between users and developers. In one sense, they could be considered the natural way to fund a free software project: just as volunteers contribute to varying degrees and in different aspects of the software development cycle, so too can users form part of the project by making a financial contribution in line with their possibilities and interests.

4. 3. 1. 4  Free product plus associated services

Companies in this category implement a strategy of the type "best knowledge here" and "best code here", developing a free product and offering services for it as a means of generating revenue.

This sections encompasses both the product specialists and badgeware of Daffara's study, since they both represent the same business model. Moreover, although badgeware licences include an additional assignment constraint, they maintain the essential characteristics of openness and freedom of knowledge and can generate the same benefits through their development communities as those that use licences without this constraint. The companies constituting examples of badgeware probably also seek to launch some sort of brand strategy, so they place special importance on assignment when redistributing the products they generate.

This model has a number of problems, such as few barriers to entry to the business – any company can gain knowledge of the product and offer services – and problems obtaining support contracts as client companies may prefer to continue with their regular service or consulting companies or to hire providers that offer support for their entire new technology infrastructure and not just for a specific product.

Another common problem faced by these models for generating revenue from services is that of innovators and enthusiasts: when a new product comes on to the market, its early users are often people with technical skills that will not contract support services for it, preferring to acquire the necessary knowledge for themselves. This model then will need to offer an extended product and transmit reliability in order to reach a potential market that will pay for services relating to the product.

The success of this type of business model is questioned by some authors (like Perens). Nonetheless, there are many companies based on this model that have attracted large sums of venture capital. For a more sustainable business model, however, they will need to address the problems mentioned above.

The models of vertical service provider specialists include Alfresco (content management), Compiere (ERP, CRM), vTiger and Openbravo.

Corporate data on Openbravo. (Obtained from http://www.camerdata.es)

Company name

Openbravo, S. L.

Head office

Pamplona (Spain)

Website

www.openbravo.com

Creation date

2001

No. of people employed in 2007

26 to 50

Turnover in 2007

Up to €300,000

Openbravo is an interesting example of this type of model. The company, founded in 2001, develops two free applications for SMEs – OpenbravoERP (enterprise resource planning) and OpenbravoPOS (point of sale) – which seek to meet the needs of management and planning and of point of sale terminals for small and medium enterprises, respectively. The code was published in 2006 and is currently among the most active projects in the SourceForge ranking.

The company has attracted substantial sums of venture capital, with investors such as Amadeus, Gimv, Adara, and SODENA (Sociedad de Desarrollo de Navarra), which has invested €5 million in the company.

Its business strategy is geared towards becoming a leading product in the industry and making OpenbravoERP the benchmark for management software among SMEs. To achieve this, the company is exploiting the possibilities of free software to the maximum through careful community management and application of the coopetition concept.

As we saw with Funambol, Openbravo observed that, alone, it did not have the capacity to disseminate and distribute its product among potential users. Although OpenbravoERP and OpenbravoPOS are aimed at SMEs rather than large corporations, in order to achieve its strategic aims of becoming the sector leader, the product had to reach countless small and medium businesses worldwide.

In addition to the size requirements for conducting a marketing campaign of this scale, the company was also aware of the potential difficulties of competing to provide services directly to end users, who might prefer local businesses or ones offering integral solutions and not just companies dealing with a single product.

To overcome these barriers, Openbravo positioned these companies as collaborators rather than competitors. Thus, it admits that, simply because it developed the product, this does not necessarily mean that it is the best company for providing related services to end users. Its mission was to create a good product that could expand markets, generating new revenue opportunities for IT services companies, which could complete its offer with OpenbravoERP and OpenbravoPOS.

Thus Openbravo defines the provision of services to other IT services companies – intermediaries between it and the end users – as its revenue generator. These companies form a network of partners that carry out the tasks of implementation of OpenbravoERP and OpenbravoPOS in SMEs.

Openbravo offers its partners various services (support, training), by implementing a pyramidal system of consulting similar to that described in module 3 of this subject, as well as conveying reliability and trustworthiness. As they are supported by the product developers, they can exploit the strategy of "best knowledge here" and "best code here" on their markets.

Openbravo thus operates a strategy of coopetition, giving service companies the opportunity to exploit Openbravo in the context of their natural markets while benefitting from the increased diffusion of its product, and obtains revenue directly from its partners. Thus far, it has been considerably successful with this strategy and currently has eighty-five partners around the world.

4. 3. 1. 5  Software as a service

Companies that develop a product can also exploit it through the paradigm of software as a service. Instead of offering installation and support services, the company is responsible for all hardware and software infrastructure, offering functionality directly through the Internet. The recurring revenue generated takes the form of service subscriptions.

A good example of this type of model can be seen in CollabNet, which provides services for collaborative software development (version control, issue tracking, communication, etc.), generated, among others, through the Subversion version control platform. In this case, in addition to keeping the source code open, the company spends a lot of effort on maintenance of the community, so that its work on the project is merely a contribution – albeit a large one – within a free community. Other examples of companies that market their products according to the "software as a service" model include SugarCRM, SocialText and JasperSoft.

With the "software as a service" format, these companies will not come across any more difficulties generating revenue than their proprietary equivalents, since the sales in this case are not derived from the copyright on the product. The fact that a client can download, install, configure, host and maintain the application will be more a tool for marketing and distribution than a loss of income. As noted earlier, corporate clients are willing to pay for having their problems solved.

Nonetheless, releasing all of the code creates problems with differentiation and opportunities for the entry of competitors. Any company with a sufficient technical capacity and infrastructure could offer a similar service if the code were available. In the light of this problem, the company that developed the product could base its differentiation on "best knowledge here" and "best code here" to gain the sympathy of the community. In addition, if its competitors also chose to contribute to the development, it could set up coopetition mechanisms, collaborating to expand the market and segmenting it later according to specialisation.

Like the mixed OSS/proprietary strategies we saw earlier, some companies in this category will implement solutions incorporating some form of restriction on their code, mainly by keeping a small section of the code closed, which will form the basis of their differentiation.