New research by industry analyst Saugatuck Technology shows that there will be a dramatic increase in the percentage of executives who plan on using software as a service (SaaS) for
business-critical operations. Gartner had predicted that by the end of 2007, 30% of small and medium business (SMB) companies would be using SaaS-based applications. This prediction is indeed
turning out to be true. So let’s explain what SaaS is all about.
SaaS (software as a service) is a software application delivery model where a software vendor develops a web-native software application and also generally hosts and operates (either independently
or through a third party) the application for use by its customers over the Internet. SaaS applications are generally available as a service for which the payment is a monthly or other recurring
fee. In most cases, that implies that there is no approval process by central IT and no wait for deployment. The model is especially appealing for small to mid-sized companies who don’t have
the skills in house to run and manage complex software applications.
Some people might confuse the ASP (application service provider) model with the SaaS model, but they are very different. The ASP model was focused primarily on providing organizations a mechanism
to offload certain application processing workloads to a third-party managed server. Also, another difference that you will find is that unlike ASP applications that were mostly client-server
programs with HTML web interfaces, today’s modern SaaS solutions are indeed designed and architected for the web environment. This sort of Web 2.0 functionality significantly improves the
usability and the level of interactivity of these SaaS applications. ASP applications were not written as net-native applications. So performance was poor and application updates were no better
than self-managed applications.
BI (business intelligence) is one of the key areas where SaaS-based models have become popular. SaaS-based applications in the world of BI are also known as “on-demand BI” applications.
As shown in Figure 1 below, the Aberdeen group report puts the future adoption of SaaS-hosted BI, on-demand BI or through an appliance at 21%. To me, it seems that the shift toward more and more BI
getting embedded within enterprise applications will fuel further growth of on-demand BI.
Figure 1: Emerging technologies and Services by Aberdeen Group
Now let’s look at 3 key trends that are definitely emerging in the world of SaaS and on-demand BI.
Mainstream SaaS Adoption: There is a huge jump in the number of companies planning to use SaaS for mission-critical applications (18% in 2006 to 49% in 2007. Source: Saugatuck
Technology). What is even more interesting to note is that there has been a surge in the acceptance of SaaS among large enterprises – enterprises with revenues of more than $1 billion. This
trend has ended the myth that SaaS is predominantly suitable for and adopted by SMB (small and medium business) companies.
Complementing On-Premise Data: Another trend that has been noticeable is the emergence of hybrid architectures where on-premise data, applications and processes are linked
through web services-based integration APIs (application programming interfaces) to SaaS-based applications. Large enterprises will probably never throw away their customized on-premise software
but will most probably look at SaaS-based applications to complement them in an effective way to improve the bottom line. And if you think about this closely, this strategy makes perfect sense as
these enterprises have invested millions of dollars customizing on-the-shelf software or building homegrown BI applications to suit their particular business needs. Some people believe that even
large enterprises will convert fully to SaaS, but I personally believe that it is a tall order and probably impossible at least in the next 3-5 years. But yes, the shift to SaaS even for large
enterprises, albeit one business function at a time, will happen.
Metrics-Based Approach: With the emergence of SaaS analytics, analytic dashboards and business performance management tools, there is an emerging trend of business services and
functions implemented using a metric-based approach. Every functional area will be driven by metrics, and they will no longer be the monopoly of business analysts from the corporate reporting
department or functional departments – people that need to spend hours and days dabbling in spreadsheets in order to run the management meetings.
In October 2007, a report was released by Gartner talked about the 5 discontinuities for IT – Software as a Service (SaaS), Consumerization, Web 2.0, Free and Open Source Software, and Global
Class Architectures. Of these discontinuities, my personal opinion is that Web 2.0 and SaaS go hand-in-hand and will probably have the most impact on IT. Web 1.0 was based on transaction,
information sharing and dynamic content. But in sharp contrast, Web 2.0 is based on high interaction, high amounts of user-directed presentation, user-integrated and user-generated applications.
On-demand BI applications have to be Web 2.0 focused in order to pique the interest of their users.
Key Characteristics of SaaS
Although there are a number of other characteristics that could potentially be listed, I have chosen to list the ones that are most significant and those that are easily understood.
Centralized feature updates: This obviates the need for downloadable patches and upgrades typical of an on-premise software installation. All customers run one version of the
base software although upgrades can also be staged for categories of customers and done in batches.
Single-instance, multi-tenant architecture: This implies a one-to-many model, although not all SaaS applications are truly multi-tenant or single-instance. A one-to-many model
implies a single physical instance with customers hosted in separate logical spaces. There can be multiple variations of how a single instance really gets implemented and how multi-tenancy really
- Managed centrally and accessed over the Internet: There is no software component installed at the customer site. All applications can be accessed remotely over the web.
Generally priced on a per-user basis: This is not the only pricing mechanism, but the most common one. The minimum number of users that companies can sign up for varies from one
SaaS vendor to another and also depends on what stage the SaaS vendor is in their evolution path as a company. Some do charge additional fees for extra bandwidth and storage.
Mostly subscription-based, no upfront license costs: This implies that functional leaders (from sales, marketing, HR and manufacturing) do not have to go through their IT
department to get them approved.
Short implementation cycle: This implies that gone are the days of the multimillion dollar Big-5 consulting projects as business leaders do not want to wait and want their
business functionality “now.”
Key Drivers Behind Adoption
Now let us look at why SaaS application models are gaining widespread acceptance.
Most people use standardized apps: Let’s look at the number of web-based calendaring, spreadsheet, and e-mail systems that have emerged in recent years. What this proves is
that an expense reporting tool, a screening tool for job seekers, a spreadsheet, or an e-mail system are all sufficiently omnipresent and well understood; and if they were asked to make a switch
from one system to another they could easily do so.
Customization and parameterization: In the world of SaaS, almost all new applications are built using business logic on top of a common application platform from parameters and
available macros. Although customization can only be done to a certain extent, there are enough options provided for the customer not to feel that they are locked into a certain way of doing
things. SaaS vendors that need code to be modified to change a workflow or report would soon become obsolete.
Reliability and popularity of web applications: There are sporadic outages and slowdowns; but on the whole, most users are willing to use the Internet for doing their day-to-day
tasks. People and companies are beginning to trustfully access remote locations and applications with low latencies and acceptable speeds.
Security is sufficiently well-trusted: Although this continues to be a challenge, the mind-set of customers has indeed changed over the last few years and they now feel much
better about data residing outside their firewall.
Improved network bandwidth: Wide area networks’ bandwidth has improved drastically following Moore’s Law – 100% increase each 24 months. Computing itself has
become a commodity.
- Low cost of ownership: Users can get up and running with just a small monthly fee and do not have to worry about any maintenance tasks nor any expensive or messy upgrades.
In the SaaS market, “on-demand BI” has been becoming popular amongst companies – big and small since 2004. On-demand BI implies business intelligence when and where you need it,
but without the hassles of building or maintaining the solution yourself, or hosting it yourself. Focus on the right business problem and clarity of understanding in terms of business benefits
become of paramount importance for on-demand BI to be successful. As shown in Figure 2 below, Aberdeen group has predicted that 21% of companies will outsource to a third party, or get SaaS enabled
as part of their respective best-in-class information delivery strategies.
Figure 2: Build it versus Buy it – Aberdeen Group
Top Challenges in On-Demand BI
On-demand BI looks promising, but it has its own set of challenges:
Data quality: Data quality is an issue with on-premise BI implementations; and so as SaaS-based BI applications gain more widespread acceptance, it will also become an issue with
the on-demand BI SaaS model. Some vendors have thought this issue through, while others do not want to consider this as part of their service. What would be interesting to see is whether SaaS
vendors really go all the way to include data cleansing as part of their service or want to direct the client to an outside provider. With hybrid architectures emerging with data shared between
on-premise and hosted applications, having consistent metadata quality across them will prove to be a challenge.
Limited customization capabilities: In the traditional world of on-premise BI, users are used to having the capability to slice and dice and essentially customize the BI
application (or, for that matter, any enterprise application) to their heart’s content. So having similar capabilities would be a challenge that SaaS vendors would have to design and
architect for in their respective products so that they can be overcome without facing customer attrition.
Scalable BI support: There would be challenges in scaling the technology for on-demand BI models, but I believe they are surmountable. This challenge is tied to the customization
capability and the data quality challenges mentioned above. The higher the degree of customization, the higher the complexity in supporting them – whether through e-mail, web or phone. Each
customer can have different data patterns and unique data cleansing issues. Explaining what the numbers mean and providing justification for them in a multi-tenant model could prove to be a huge
Security: Although the customers have generally accepted the fact that data can safely reside outside their firewall, concerns around this area will continue to get raised as
more and more business functions get moved over to the SaaS model. I feel less worried about this as “effective management of customers’ data” in a multi-tenant model is a
virtue of necessity and a question of survival for the SaaS vendors. So there are no options not to do this right as no vendor would want to go into oblivion.
What Does the Future Hold?
In the next 4-6 years, it would not be surprising if one sees a significant portion of new business software getting delivered through the SaaS delivery model. Most companies would have deployed at
least one SaaS application, and SaaS would be a key component of enterprise architecture. In 2010, Google alone will consume 30% of all the world’s servers, so global computing will no longer
be just a phenomenon. The top two challenges for adoption of on-demand BI, as predicted by Gartner, are security and scalability. With demand increasing, scalability would be something that the
SaaS vendors would have to grapple with. According to the Gartner report – “CRM Beyond Web 2.0” in 2006 – it has been predicted that through 2012, 20% of CRM applications
will blend predictive analytics and operational CRM into one system. It is my personal belief that predictive analytics and the possibilities that could emanate from leveraging technologies like
data mining and predictive modeling would be the most interesting area in which innovative SaaS-based applications will get introduced as enterprises (SMB and large ones) look for new ways and
means to improve their bottom lines and stay ahead of competition in a slow economy.
Next in the Series
This is the first of a series of articles on SaaS and BI. The next 3 articles to be published in the September’ 08, December’ 08 and March ’09 editions of TDAN will cover the
On-Demand BI versus On-premise BI: This article will present a detailed analysis of how SaaS and on-demand BI will affect the traditional world of on-premise BI. From a
practitioner’s standpoint, we will explore the situations in which using SaaS-based BI would be more advantageous and when using a hybrid approach would be much more beneficial. We will
also discuss SaaS architecture models as they apply to the world of on-demand BI.
Should IT embrace SaaS: In this article, we will explore Gartner’s 5 discontinuities for IT in depth and understand how they impact the psyche of IT management. We will
talk about some ways for IT to deal with the upcoming trends in SaaS and the 5 discontinuities predicted by Gartner. Software vendors will be going around IT to gain adoption of their software,
so IT will have to stay alert. The article will provide some tips and techniques to deal with this.
Impact of SaaS on DQ (Data Quality): The impact of SaaS and on-demand BI is going to touch the world of data quality too as DQ becomes more and more important (almost
mission-critical) for BI projects to be successful. We will talk about what is currently happening in the DQ marketplace and explore the upcoming trends and predictions.