We know intuitively that complexity drives inefficiency and cost, so it's no surprise that organizations that don't let their enterprise applications proliferate like kudzu perform better than peers that take a less controlled approach to application management. In its Information Technology Metric of the Month, the Hackett Group finds that organizations it classifies as world-class have 20 applications per 1,000 end users, while more average organizations have nearly twice as many, 39 applications per 1,000 end users. (You can download the metric on Hackett's website.)
Number of applications is a "reliable proxy" for the complexity of technology supporting an organization's business processes, according to the Hackett Group. Tending too many applications hampers IT's ability to respond quickly to business requests and to pursue more strategic tasks. And complexity breeds complexity. Companies with lots of applications also tend to have more technical platforms, more data models, more fixes and upgrades, greater fragmentation of expertise, and more duplication of work, all of which tend to slow project completion.
Honorio Padron, the Hackett Group's Global IT Advisory Practice Leader, told me a lack of governance is one of the key causes of overextended application portfolios. Too often, he said, companies focus narrowly on budget and specifications during IT projects and neglect to shutter applications that may no longer be needed following the implementation of newer technology. Companies need a more holistic analysis of their entire application portfolio, one of the key tenets of project portfolio management. Yet The Hackett Group finds only about half of companies consistently use such an approach, Padron said.
Not seeing the big technology picture is one of several project portfolio management mistakes I wrote about in a February post.
Even companies committed to standardizing their technology platforms as much as possible and weeding out duplicative or unnecessary apps may face pushback, Padron added:
It's like a restaurant that adds a new menu item. When it goes to drop an item to make room for the new one, somebody probably says, "You can't get rid of that. Three customers in the world still eat it."
The first step for companies looking to cull apps is to establish a baseline, said Padron. After doing that, they can begin identifying likely candidates to send into the sunset. Some obvious indicators:
Companies should perform regular inventories of their applications, Padron suggested. Doing so will not only help target applications that can be eliminated, but can yield other savings opportunities such as reducing the number of software licenses.
IT departments sometimes take advantage of circumstances to promote a project portfolio-management approach. Padron said companies closely evaluated their applications in the lead-up to Y2K, welcoming the opportunity to jettison apps rather than rewrite code, and following the passage of Sarbanes-Oxley, when mandated transparency helped identify apps that could be retired.
In the same way, the recent economic crisis prompted companies to look at their portfolios to determine where they could trim unnecessary apps to cut costs. But with the economy on the mend, it's not clear whether companies will make fundamental changes in their IT policies or go back to business as usual.
From the Hackett Group's Monthly Metric:
As time goes on and the most immediate effects of the recession begin to fade, companies will face a different kind of test: how well they adjust to a new economic environment, one characterized by higher levels of volatility in areas such as demand, commodity prices, exchange rates and the regulatory environment. Since IT is so intrinsically linked to operational and customer-facing processes, as well as other [general and administrative] processes, it is essential that it be as effective as possible. This starts with minimizing application complexity. Reducing the applications per 1,000 end-users increases IT's effectiveness, which in turn increases the efficiency of the other functions.