"But you end up with this huge application that is never going to finish on time or on budget. The CEO and CFO look at this financial disaster and realise the app performs badly and has cost a lot of money."
Purchasing decisions need to shift to consider operational performance metrics beyond features and functionality, and how new SaaS apps will contribute to the way the business runs in the future, Carlton says.
The IT organisation also needs to provide limitations to the business around the purchasing of cloud solutions. CIOs must drive home that data must be kept in a certain format, for example, and that interfaces to the data comply with open industry standards, he says.
Blindly selecting SaaS vendors that don't provide access to your data is another mistake, says Bhatia. This creates problems when people want to do data analytics to optimise their business processes.
"Having access to your own data is key -- the only way you can make a data warehouse work is if you have data from disparate sources in the one location where you can make queries across systems," Bhatia says.
This was particularly important when Carsales was dealing with its secondary SMS provider.
"This provider implemented the same API as our primary provider had for us so our integration costs were minimal," Bhatia adds.
"The majority of work with SaaS applications is in integration. If you pick providers that make it hard to integrate for whatever reason, it could be technology or it could be they are too precious about how they want to do business."
Sign up for Computerworld eNewsletters.