Gartner published a report this week on the "five most-common SaaS assumptions." I agree with all of them, but must throw in my own two cents on the first assumption: SaaS is less expensive than on-premises software.Gartner says that assumption is true during the first two years, but not for a five-year total cost of ownership. Well, I say, sort of.

I've been trying to get at the bottom of SaaS vs. on-premises software costs in recent weeks. So I recently picked up the phone and sought the opinion of a CIO I know who uses both: Manjit Singh of Chiquita Brands. For example, he has Oracle/JD Edwards on premises, and uses Workday's human-capital management SaaS.

Singh acknowledges that the answer of whether SaaS is cheaper depends on the time frame. Anything beyond five years, he said, and licensed software appears to be a better investment.

Still, Singh said, "I would argue in most cases we would update our software every three to five years." In some instances it's a "forced migration," meaning the on-premises software vendor makes it difficult to stay on an older version of the software if it wants to move its customer base to something new. In other cases Chiquita might want to make a change to a better vendor or product.

Singh says he's specifically studied a three-year period of costs associated with SaaS vs. on-premises software, and in every instance, "we're either better off with SaaS or it's a break even." That's just on the pure software cost side, he said; it doesn't include what he would pay to hire consultants to integrate or customize on-premises software.

There's no question about it: When CIOs are considering SaaS vs. on-premises, how long they plan to have the software should be one of the most important factors in the decision.