Software as a Service Is Gaining Ground


One does not have to closely watch the enterprise applications market to realize that the hosted delivery model is enjoying a new glorified, (or reinvented, if you prefer) status. Referred to as on-demand, utility computing, or software as a service (SaaS) delivery approaches, hosting has not only achieved buzzword status overnight, but the concept has been gaining ground through real deployments. Hosting was once known as application service provider (ASP), which has negative connotations stemming from the " propaganda, boom and bust," but this term is being replaced by these other, "sexier" terms (which are also quite possibly better value propositions). Industry giants (and some thought-leaders, which are not necessarily large vendors yet) are spending significant marketing dollars vouching for this old concept reborn.

Although there are subtleties and even distinct differences separating these terms and their associated business models, the various hosting flavors are all variations of the same market. New business models, markets, and providers are taking advantage of Internet-based technologies and standards to provide solutions based on the notions of standardization, interoperability, software component reuse, and automation. These terms describe a shift away from traditionally heavily customized (and supposedly unique) packaged software suites, owned and managed by user enterprises, which were expensive and time-consuming to develop, implement, and maintain (see The "Joy" of Enterprise Systems Implementations), and toward standardized, componentized, common, and lower cost software services sourced (and even cancelled) at will from service providers.

Whether referred to as hosted services, ASP services, SaaS, utility computing, or software on-demand, the idea is basically the same: instead of buying and installing expensive and pesky packaged enterprise applications, users can simply access applications over a network, with an Internet browser being the only absolute necessity. Thus, often there is no software and hardware to buy per se, since the application is used over the Internet and is paid for through a subscription or supported by a third party, such as an advertiser. Advertising-based software offerings emerged several years ago in the form of Web e-mail and Web calendaring. More recently, advertisers have realized that Web-based software applications are just another type of digital content that can be used to reach a targeted audience, and companies such as Google, Yahoo!, and Microsoft, are refining their capabilities to target users with context-based advertisements.

Shift to Hosted Models

Burned by negative experiences deploying unwieldy packaged software suites on their premises, users are now wiser and more assertive. Increasingly, they are demanding software that is easy to purchase, easy to consume, and has tangible payback or return on investment (ROI). For users, their purchasing demands have almost become akin to retail purchases that can be made via credit cards. Rather than spending millions of dollars on software before seeing any inkling of ROI, users have started to demand a performance-based, shared risk model of software provision. The licensing and delivery of enterprise software products is therefore undergoing a fundamental shift from traditional upfront fees for perpetual licenses to incremental, per transaction, pay as you go (PAYG), or even success-based pricing. These are becoming popular alternatives, especially for small businesses and startups that do not have the same, large information technology (IT) budgets as larger, established companies. Companies can acquire software for a lower entry cost and pay for more only as their business expands (see Trends in Delivery and Pricing Models for Enterprise Applications: Pricing Options).

Though the price of hardware is decreasing and becoming more affordable, a major predicament remains: enterprise software applications, such as complex enterprise resource planning (ERP), supply chain management (SCM), product lifecycle management (PLM), and customer relationship management (CRM) systems, are often too expensive and too intricate for small companies to govern. While they can afford the necessary hardware, they do not have the IT staff and infrastructure required to support a major enterprise application. Relatively cheaper solutions are now becoming widely available, and vendors are addressing their customers' desire to use technology as needed. Users do not want to buy entire software packages or infrastructure when, typically, only a small percentage of the overall capability is used. User enterprises have also become more agile, requiring more flexibility in IT delivery and usage, as well as licensing and payment structures, and vendor business models. In the enterprise applications space, many customers are moving away from large upfront licensing contracts (with ongoing maintenance fees), to one that is variable, based on usage, and defined by a subscription-based relationship bundling the entire offering. There might also be a financial advantage to having the software be an expense rather than capital, which will depreciate over time.

A slew of recent moves by the most prominent contemporary market players may be another sign of enterprise applications' slow but ongoing evolution to SaaS, on-demand, and related hosted models. The market may be experiencing the beginning of the end of traditional user-based licensing on the customer's premises for a given period of time and product version. The vast majority of business application software vendors still generate most of their revenues by selling their software licenses (via compact discs [CDs] or similar physical media gadgets) based on the number of named or concurrent users or seats or on the number of processing hardware units (and sometimes on a per module basis, though this is often based on an unnecessary "wall-to-wall", "all you can eat" functional scope). Revenues are also derived from accompanying implementation services, post-implementation service, and support and maintenance, which are priced as a percentage (or more often as multiples) of these software license fees.

This model might be wearing out its welcome on both the vendor and customer side. For one, it tends to be cyclical, since vendors first sell their present product versions into the market, and then sell subsequent upgrades. Logically, sales revenue should peak after each major upgrade or product release, and then drop until the next one (on average every twelve to eighteen months). This creates a cyclical, yet erratic revenue stream, which, in turn, creates cyclical, volatile stock prices (for public vendors) and also has other business performance-related ramifications. Bundled with this is the inability of licensed, on-premise, packaged software to keep up with ever-evolving "best industry practices", since traditional packaged software is largely stuck in revolving major release cycles. These cycles require most of the research and development (R&D) effort to be spent on building and testing compatibility for operating systems (OS), databases, application servers, and other platforms.

On the other hand, many user companies are unhappy with the rigidity of the model, especially in terms of the tiresome and endless upgrade process, and maintenance fees that keep creeping up. This, coupled with the non-standard pricing, leaves them wondering what kind of deal they have gotten (and particularly whether they are paying for functionality they will not use any time soon). To put this into perspective: how often do we buy a lifetime's worth of snacks and coffee during one trip to the grocery store? Such shelfware comes in many forms, such as products that are acquired but never used, or modules that are bought as part of an entire suite. They are modules that were once bought to fulfill a function that no longer exists in the business, for whatever reason; and capabilities that have become redundant as they are replaced by new software applications. Though these gratuitous capabilities are hardly ever used, they still require license and maintenance fees. SaaS gives users the option of buying software applications as appetizers, which are far more sensible portions that satisfy a need. It decreases software bloat, and is far more kind to the IT system's "waistline". For more information, see Application Erosion: Eating Away at Your Hard Earned Value .

The widespread use of personal computers (PC), the Internet, and ensuing Web-based applications has had an essential impact on the way business applications are being sold and delivered. The development of Web-based applications has decoupled the user interface (UI) from the business application logic and its underlying software and hardware platforms. As a result, any user working at a PC with a Web browser can access a variety of business applications running on different software or hardware platforms in any number of different locations for the cost of a mere Internet connection. This decoupling has therefore allowed enterprise application vendors to begin to change their business model from traditional on-premise software license sales to the delivery of SaaS offerings.

However, a more important factor in the shift to SaaS might be a change in the way the software itself is created nowadays. Rather than software components being developed and bundled together to form a monolithic, rigid solution, systems are increasingly being developed as a "federation", "mash up" of services, or composite applications, which are only tied together at the point of execution. This will eventually enable alternative software components to be substituted between each use of a system, allowing much finer grained flexibility. A simple analogy is the use of an electrical appliance. The user does not directly negotiate with the electricity company to use power for the specific appliance. There are standards and controls, but they are broad enough that an electrical appliance can be plugged into the service without the user's notifying the electrical utility. On its side, the electrical utility takes care of the complexity of power generation, including matching capacity to demand, and it can change which generators and circuits deliver the power—all without coordinating these events with the millions of users who rely on the service.

Hosting Evolution or More Intelligent (Software) Design?

Applications hosting was the model where another company ran the software and hardware for a user company, which may or may not have owned the software license. A reinvention of the old service bureaus of a few decades ago, customers paid a fixed fee for a service and rented access to the software over the Internet or a virtual private network (VPN). The hosting party, called an ASP, could be the software vendor or another third party entity that managed and distributed software-based services and solutions to customers across a wide area network (WAN) from a central data center. In essence, ASPs are a way for companies to outsource some or almost all aspects of their IT needs. They may be commercial ventures that cater to customers, or even not-for-profit or government organizations, providing service and support to users. ASPs emerged on the Internet in the late 1990s in response to the expenses, upgrades, and technical difficulties, including misbehaving applications, random server downtimes, and compliance issues suffered by enterprises.

But, while ASPs eliminated many application deployment and maintenance issues, the ASPs and hosting companies were unable to substantially streamline the traditional packaged software they offered, and did not make the software easier to access and use. Moreover, most of the first generation of hosted applications was not suitable for delivery over the Internet and certainly was not appropriate for the small and medium businesses (SMB) for which the ASP or hosting model was regarded as most appropriate.

For example, traditional ASPs required retailer and manufacturing user enterprises to predict their peak load based on past performance and forecast speculation, generally in a short selling season that required a specific number of servers for a short period of time. By doing so, there were significant and unnecessary expenses and long-term contracts that locked the enterprise into the maximum potential processing capacity, which sat inactive for most of the year. On the other hand, if business performance was better than expected, no additional servers would be available and the e-commerce site would experience slow performance at the peril of the retail business.

The main challenge that faced most ASPs was how to drive down long-term costs while accumulating a solid revenue stream. The earlier wave of hosted offerings, based on using a third party client/server system, lacked economies of scale. Thus, most of these ASP and hosting industry pioneers disappeared because they could not convince enough mainstream enterprise customers to replace their homegrown or packaged applications. Simply put, they could not generate enough revenue to offset the tremendous upfront capital investments needed to build and maintain hosting facilities. One of the cost inhibitors for ASPs was the amount of dedicated bandwidth they had to maintain to support thousands of users. Another challenge was service level agreements (SLA). If the ASP lost Internet connectivity, then customers would also lose connectivity, which would negatively impact their internal SLAs. To make things worse, many of these "flash in the pan" companies went public before making any profits, when investors were already disillusioned with the bubble.

Another problem was that ASPs over-promised their ability to host heavily customized enterprise applications environments. Their commitments were not technologically practical at the time, and they often tried to fit a "square peg into a round hole" by hosting applications that were simply not amenable to hosting at that time. The results were widespread failures, the mass demise of early ASPs, and the disgrace of an otherwise great idea as a positively dull, reliable, service-oriented business model. In 2002, ASPs were pronounced as an all but failed business model following many real life disasters and the availability, privacy, security, compliance, control, customization, and cost concerns of customers and market observers that ensued. For more information see Hosting Horrors!.

In fact, the term "ASP" was (and still might be) confusing, or incorrectly used. The hosting model involved vendors that were really application aggregators, rather than software developers. These aggregators provided only on-line versions of on-premise licensed and packaged systems. Their approach was more focused on outsourcing and hosting than on providing a software application as a service, even though the evolving requirements of business had long since dictated a move away from this model.

Thus, ASP and outsourcing are two distinct models. Traditional software vendors know little about the intricacies of networking and hosting, as their expertise lies in software, so there is still room for true ASPs. These ASPs should have little to do with outsourcing or hosting, but rather should provide SaaS. The way in which application service provisions are defined and deployed is being revisited. Emphasis is increasingly placed on applications and services, whereby, in their quest to become SaaS providers, ASPs need to do away with the traditional perpetual software license model. There should be no talk of product, users, or price; neither should there be any up-front payment nor annual maintenance fees. Rather, the focus should be on value, which is defined in relation to the business value that the customer or user derives from the use of the software service. Hosted applications can be purchased on a "rental" basis. This means that instead of buying a software license, customers pay a subscription price per user per month (or year), or per number of transactions in certain software categories.

ASPs that have reinvented themselves as SaaS providers have done so by first focusing on one or two core areas, such as sales force automation (SFA), e-mail management, e-commerce storefronts and product catalogs, human resources (HR) or payroll, or financial software. They also offer only their own software, written from ground up entirely for an Internet-based architecture, and run the software on shared servers to minimize costs. This is the next step in the evolution of hosting services.

This concludes Part One of a four-part series, Software as a Service Is Gaining Ground. Part Two will examine key features of SaaS, while Parts Three and Four will look at SaaS vendors and provide user recommendations.

comments powered by Disqus