As cloud computing is still very much at the peak of the Gartner Hype Cycle (and very much on the mind of CFOs), it’s worth examining how cloud computing can add both complexity and cost to what should be a relatively standard process: selecting an enterprise IT system in a midsize organization. Sometimes, coming down from the clouds to look at a real-world case is a great way to focus the mind.
Case in Point
A few months ago, a midsize Australian organization decided it needed to either upgrade or replace one of its legacy enterprise applications. This organization, with about 3,500 employees, has upward of 20,000 system users across its customer base. The organization approached a number of local and global suppliers, and some responded with on-premise offerings (meaning the application would be loaded onto the business’s servers), while two responded with software-as-as-service (SaaS) solutions (in which the application would be hosted by the provider and accessed through a browser). Surprisingly (to the organization, at least), what was expected to be a relatively straightforward exercise was not, due to the nature of SaaS pricing.
The first thing the organization’s review panel tackled was the challenge of comparing costs between an on-premise application and one hosted in the cloud. The costs of both cloud systems were based on a blend of per user, per month license subscriptions, plus a per transaction cost for certain transaction types.
Based on the business’s current transaction volumes, the initial cost analysis showed that at the end of 24 months, the total SaaS costs would be between 135% and 280% higher than the on-premise equivalents. If the analysis were extended five years (the projected life expectancy of the application), the difference widened to between 280% and 350%. Those costs would be significantly higher than the expense of hiring additional IT staff to support the on-premise applications. Factoring in upside transaction growth projections (based on the organization’s strategic plan) further increased the SaaS cost over the on-premise alternatives by an even wider margin.
The on-premise solutions the organization considered had fixed license costs and fixed annual maintenance charges with no exposure to escalating transaction volume costs. In addition, once the initial costs of implementing the on-premise product were depreciated, the cost benefits of the product as compared to the SaaS solutions became even more apparent after year three.
The key driver for this large differential between the cost of the SaaS and on-premise products was the transaction cost escalators that were based on usage volumes. The per user SaaS subscription model worked fine for internal users (full-time employees) because the organization could control their use of the application. However, this application needed to be exposed to the external client base, and the organization’s ability to limit self-registrations and other usage patterns by that group would be difficult, if not impossible. If one could take this variable — the per transaction cost — out of the SaaS equation, the cost differential would be eliminated. The discussion could then move to questions of functionality, performance, and so on, rather than be stuck on cost alone. But without removing the per transaction cost, it’s like comparing chalk and cheese.
Making the Simple Complex
Cost considerations aside, the SaaS offerings introduced additional complexities. Of primary concern to the purchasing organization was the need to deploy a small number of third-party cloud solutions as a bolt-on to the core offering to handle system-to-system integration, identity management, access control, and so on. More players in the overall solution introduced additional cost and risk. If, for example, any one of these ancillary providers went out of business, the operational integrity of the overall system could be compromised. Moreover, governance around software change control could also be further complicated, as keeping the various systems working together as a whole when one is upgraded would require careful orchestration.
The combination of a cost exposure plus governance complexities may not be a good mix: something that might keep CFOs awake at night.
Lessons Learned
- If your cloud provider is charging on a per transaction basis, make sure you model your costs under various usage patterns and volumes for the expected lifespan of the system.
- Do not make the cloud your default position; test any assumptions.
- If you are not fully up to speed with the cloud paradigm, get a trusted, truly disinterested and experienced partner (one who has no financial interest in the outcome) to help guide you through the assessment process.
- Do not underestimate the cost, complexity, and risks of integrating a cloud offering with any of your other systems, on-premise or otherwise.
- Insist on getting a price for features and functions that you may need in the future.
Of course, there are exceptions to every rule. Just make sure that when you launch your own cloud initiative, the driver is not enthusiasm for the technology but a deep knowledge of your own business requirements.
Former CIO Rob Livingstone is an author, speaker, academic, and consultant with substantial real-world cloud experience. Join Rob at one of his many free live webinars at www.navigatingthroughthecloud.com.