Is the SaaS experiment finally over?

Software as a service (SaaS) has been one of the most important trends in software development in recent years, particularly for the small and midsized business markets. As the sales pitch goes, SaaS offers customers streamlined licensing and billing while reducing costs associated with infrastructure, support, and maintenance. But according to new research publicized by Gartner, the pitch might not match reality for many customers.

At a presentation at the recent Gartner SOA & Application Development and Integration Summit in London, Gartner analysts said SaaS “will have a role in the future of IT, but not the dominant future that was first thought.” Indeed, for all the hype around SaaS it has hardly taken the software industry by storm. By Gartner’s estimates, SaaS as a percentage of total enterprise spending grew by only 0.6 percent from 2008 to 2009. That should give software developers pause, particularly those who are only now planning to roll out a new SaaS offering.

Customers question the value of SaaS Naturally, cost ranks high among customer concerns. In 2009, Gartner published research that suggested vendor claims about the cost effectiveness of SaaS versus on-premise software may be inflated. According to Gartner’s figures, SaaS can be cheaper in the short term, but over five years, the total cost of ownership of a SaaS deployment can actually be higher than the cost of the equivalent on-premise software.

Budget issues aside, customers who are considering SaaS for mission-critical applications have further concerns. Chief among these is whether SaaS vendors will truly be able to deliver the uptime performance they promise. Many customers still feel more comfortable relying on their own IT staff to manage and maintain on-premise software, rather than putting their complete faith in a SaaS provider.

And with good reason — doing SaaS well is difficult, particularly for a large volume of customers. Even the most prominent players have experienced outages. Early on, pioneering SaaS provider Salesforce.com experienced repeated blackouts. And in 2009, an outage at Google’s popular Gmail hosted email service was attributed to overloaded servers, despite the company’s massive server infrastructure.

Software developers who are readying SaaS applications of their own should take note. Such service lapses are virtually impossible to avoid, and they can bite SaaS providers even when their software itself isn’t to blame. Last year, leading hosting provider Rackspace suffered three major outages, each of which knocked thousands of applications offline. Meanwhile, Google, having rolled out a business-oriented version of its App Engine cloud hosting service in May, has struggled to keep up with the CPU demands of its data store, resulting in poor performance for many customers.

Gartner’s advice to customers: Tread carefully The bigger picture for developers is that a customer’s relationship with a SaaS provider is markedly different than its relationship with a traditional software vendor. While purchasers of on-premise software typically evaluate vendors on such factors as cost, scalability, ease-of-use, security, upgrade path, and licensing model, SaaS customers have to weigh another set of considerations. “SaaS changes the role of IT from implementing its own operations to inspecting a vendor’s operations,” says Gartner vice president David Cearley.

Gartner advises its clients to perform extensive diligence before signing with any SaaS vendor. That includes not just weighing the costs and benefits of a specific solution, but also developing an in-house SaaS governance policy to help gauge the solution’s real-world performance. Such a policy should be a collaborative effort between business and IT, Gartner says, and it should consider not just the business performance of a given SaaS vendor, but its technical and operational capabilities as well. That means SaaS vendors will need to be transparent enough in their operations to instill customer confidence in their offerings.

The ultimate worst-case scenario, according to Gartner, is when SaaS becomes “shelfware” — software that is bought and paid for, but never used. Maybe the customer had a round of layoffs that resulted in vacant seats, or maybe it over-purchased in anticipation of growth or in hopes of securing a deeper discount. Either way, that customer isn’t realizing the full value of the product.

This is a common enough occurrence even with on-premise software, but it’s especially risky for SaaS vendors. Because SaaS uses a pay-as-you-go model, a customer audit that turns up shelfware can result in unexpected large revenue losses for SaaS vendors, virtually overnight. As a result, SaaS vendors must be especially vigilant against over-selling and should monitor usage patterns to ensure that their customers are getting the most out of their products.

What’s your business: Software or IT? The bottom line is that software developers who are considering offering a hosted version of their products are pondering a very different business than traditional software development. In addition to the usual developer concerns about code quality, debugging, refactoring, feature release cycles, and so on, SaaS vendors have to worry about provisioning network bandwidth, securing running servers, scaling server infrastructure (both for growth and for demand spikes), backing up customer data, protecting user privacy, and more.

While these issues may seem obvious, it’s worth considering that even when all this is taken into account, customers may still be leery of SaaS for many applications. According to Gartner, SaaS has been most successful in the content management, collaboration, and CRM markets. Whether enterprising SaaS vendors will succeed in pushing the model into new markets is a matter of conjecture, but based on current growth figures, it sounds like an uphill battle.

The question for software vendors, then, is whether it really makes sense to venture into areas that might not fit their core competencies, given the current climate. Perhaps a better idea would be to experiment with new pricing, licensing, and service models to make on-premise software more attractive to those customers who may be on the fence. Commercial open source is one example of an innovative on-premise software business model. Are there others?

This article, “Is the SaaS experiment over?,” originally appeared at InfoWorld.com. Read more of Neil McAllister’s Fatal Exception blog and follow the latest news in software development at InfoWorld.com.