By Emily Chia | Jan 14, 2011
CloudGarage.Asia founder and CEO Mark Ross will not be forgetting his first foray into the cloud. After experiencing the immense gains of software as a service (SaaS) at SunLife, which cost a third of traditional hosted software and was four times faster to roll out than traditional software, he soon decided that could computing was the future.
Formerly CIO of financial institutions AIG and Sunlife, Ross' SaaS journey started in early 2007. Sunlife, the organization he was in, had presence throughout Asia but the service offered was inconsistent.
In most cases, the service solutions were manual and may involve relatively expensive calls to the call centers. The unscripted customer interaction resulted in inconsistent service experience.
"Most importantly, there were few cases of less expensive online customer service across the region. But our main aim was to create a consistent customer experience--one in which the customer had options, for example, to be serviced not just via a call center but also through self service online."
The other aim was to reduce the total cost of service (TCOS).
He estimates that it costs five times more to satisfy a service request via call center compared to online service transactions.
All this happened during a period of rapid Asia business growth. In view of the financial and human resources constrains, the "e-services" program was conceived, said Ross.
With the e-service program, the organization aimed to offer customer self service, establish a consistent experience, improve customer loyalty, reinforce the brand, reduce operational expenses, deepen the customer relationship, and establish a stepping stone to e-sales.
Delivering promises
Ross eventually decided to use SaaS to deliver the online service he envisioned.
Challenges were plentiful, but his perseverance paid off. SaaS enabled the service to be consistent. In addition, delivery cost was reduced to about 1/3 the cost of using a traditional architecture which included the hardware, software, data centre space, staffing, maintaining of the operating systems and data backups.
Customer access and service improved, he said. Non-technical issues, including negotiating pricing and organizing budgets; preparing the technology architecture, and finalizing the business case, required eight months to complete, while development required four months.
Implementation in subsequent countries required two months, he added.
Challenges faced
He remembers the many constraints he faced.
"We needed one common solution for all of Asia. Rapid prototyping and time to market was important too."
He wanted to minimize demands for new staff and also maintained the service centrally and cost effectively.
He recalled that while the business growth rate for SunLife in Asia was strong, the businesses were relatively small. Thus Asia's financial and staffing budgets were limited.
For this reason, the technology organization was severely constrained on the number of new hires for this initiative.
"Interestingly, once the direction was proven and successfully delivered, senior management was more relaxed about investing in incremental resources, including staffing, to support the initiative."
In addition to minimizing demand on constrained data centers, security and scalability were his top priorities too.
There were many SaaS risks anticipated, he said. To counter security issues, he conducted deep physical and logical security audits of SaaS providers. To ensure performance, he leveraged prototypes, SLAs, customer references and performance statistics. And the organization's SaaS inexperience was countered by establishing partnerships and internal CoE; organizing training; and starting implementation on a small scale.
When Ross linked the SaaS to proprietary legacy systems, the issues he faced went beyond cleaning the data.
"We needed to go through an exercise of understanding our data - what was the definition/meaning of the fields, what was the data architecture."
Some of the information was hard to come by for the legacy systems involved, said Ross.
"The lesson to be learned: assume that you do not know your data before you start--and you will be less likely to underestimate the time required to prepare it for SaaS. The data management stage of the project was included in the eight months preparation stage and it took about three to four months of the eight."
And as it was his organization's first SaaS initiative globally, he had to assuage HQ concerns on issues including legal, and regulatory.
He remembered that the first SaaS architecture failed to sync ground-based and cloud-based data sources within time constraints.
Fortunately, the impact on schedule was minimal as the milestone of creating a working prototype was incorporated into the plan. It was more a psychological impact, said Ross, as the first design took 18 hours to complete the desired upload when it should have taken less than two."
"We got nervous. We huddled around the problem much in the same way that Mission Control huddled around the Apollo 13 problem."
Eventually, his team solved the problem by repositioning the role of the Extract, Transform, and Load (ETL) appliance (focusing it only on data transformation and not on data transformation and storage), establishing multiple parallel processes to work on the data upload simultaneously, resolving malformed data issues that were stopping the upload process, establishing operational procedures to be responsive to any process slowdowns.
At that point, the SaaS commercial models were also unprepared for the solution to be used by external versus internal stakeholders.
The service portal, he explained was to be used by thousands of external stake holders. The license models at the time only accounted for the use of handfuls of internal users.
"Our first Total Cost of Ownership estimate was completely impractical, as the vendors gave us a cost/internal seat price to estimate the cost of each customer using the portal. This metric resulted in a financially impractical solution."
He sat down with the vendors and explained that SunLife's approach was different and they needed to reconsider their pricing models. He is happy to say that this issue has since been understood by the vendors and they have appropriately adjusted.
With active participation by Philippines, New Zealand, India, Ireland, China, USA, Canada, Indonesia teams, he realized he had to create a SaaS Project Management Office (PMO)
Ross was aware of this situation from the start, as he had started the SaaS undertaking with smaller and more manageable steps and projects. Thus, the SaaS PMO was built in advance of the e-services initiative and was indispensable in helping to coordinate and manage the complexities of the SaaS program.
The good thing about having program metrics in real time and available to all was that it helped to quickly illuminate trouble spots, he reflected. As a result, the program team was able to efficiently focus its resources on the priorities and that increased the success of the outcome.
0 comments:
Post a Comment