Software-as-a-Service: The Dirty Little Secrets Of SaaS

The IT industry may be embracing the model of Software-as-a-Service (SaaS), but there's just one problem: Up to 20% of attempted SaaS deployments are failing due to serious problems with data integration. And that's just one of many problems facing the hot but problematic movement.

This fact doesn't make the headlines, as vendor after vendor instead hails the signing of marquee clients for their particular service. But after the media spotlight fades, new SaaS customers are often left with shiny new portals for an online service, and plenty of headaches trying to get their legacy data to work with the new service.

That enterprises and other businesses are attracted to SaaS is nothing new: a 2010 Forrester report commissioned by Symantec reported even then that "58% of enterprises use two or more SaaS-based business applications today, and 72% plan to in 12 months. More strikingly, 19% of enterprises report having six or more SaaS-based business applications today, and 30% plan to have that many in 12 months."

There is little evidence that this situation has changed in the past 27 months, according to Mike Hoskins, CTO of Pervasive Software. Indeed, as SaaS technologies have improved, applications have become even more robust and easier to deploy.

Integration Is Still An Issue

"But integration is still an issue," Hoskins said. Even in the 2010 report, 39% of SaaS customers reported concerns with data integration - a concern second only to security worries. Today, the outcome of many SaaS deployments paints an even bleaker picture.

"49% of our potential customers report difficulties with post-deployment integration," said Lance Speck, General Manager, Integration Products at Pervasive. And these are not trivial problems, either, Speck emphasized. They are enough to derail the entire deployment operation for around 20% of SaaS migrations.

It's not just the potential for failure. The costs of data integration can be huge, and this potential expense is not typically mentioned by SaaS sales teams knocking on enterprise IT's door. For every dollar spent on a customer resource management deployment," Hoskins cautioned, "a customer can spend up to $5 on integration."

These figures punch a big whole in the theory that SaaS is the end of the rainbow for enterprise IT. And data integration is not the only problem facing SaaS deployments.

Not Just Data Integration

Security remains the prime concern for SaaS deployments. Actually, it was a huge concern, and now it probably ranks high on the "Something Wicked This Way Comes" scale, as service after service this year have announced hacks that have compromised their security.

(See also Evernote Is Latest Hacking Victim and World War III Is Already Here - And We're Losing for more on SaaS security.) 

Data portability is another pitfall of SaaS deployment. If you spend all that time and money getting your data to work with a particular service, the last thing you want to hear is that this service is about to shutter its doors. We've seen quite a few consumer services roll up their welcome mats, enough to warrant concern for enterprise customers.

(See also Sudden Site Shutdowns And The Perils Of Living Our Lives Online.)

It's not just the threat of a SaaS vendor dying, either. Price changes, poor service or a better mousetrap are all valid reasons for wanting to pick up your data and move. With integration so expensive, that makes choosing the right SaaS vendor all the more important.

Warning Signs

Speck warns IT customers to be on the lookout for tell-tale signs that may indicate that a SaaS vendor may be trying very hard to get that initial signature on the contract - and not worrying about the all-important follow through.

"Whenever the SaaS vendors says, 'we'll worry about integration in Phase II'," it's a big red flag, Speck said. Another call from the cluephone? "'Don't worry about integration, we have an API,'" Speck added.

APIs (application programming interfaces), which enable a customer's applications to talk to the service's code, are indeed an critical component of data integration, but the mere existence of an API is only a part of what's needed. Other relevant questions to consider: How open is that API? How well documented? And how easy is the API is to use?

Not every SaaS vendor should be greeted with suspicion, of course, but enterprise IT departments should consider the entire process, from deployment to integration, as well as security and data portability, when selecting a SaaS vendor.

 

Image courtesy of Shutterstock.