It’s an adage as old as time (or at least as old as the invention of the personal computer): Technology is destined to cycle constantly between complexity and simplicity.
Remember the hassle of attaching peripherals in the days before USB ports? Remember the anguish of developing applications for competing OS interfaces before HTML? We fixed those problems, and look at that, we’ve moved on to others.
“Complexity grows over time,” says Bryson Koehler, chief information and technology officer (CITO) of The Weather Company in Atlanta. “Systems are built to do one thing, and then they’re modified, morphed and bastardised to do things they were never meant to do.”
Complexity also occurs when technologies overlap one another – “when you add new stuff but keep the old instead of getting rid of it,” says Dee Burger, North America CEO of Capgemini Consulting.
Even as recently as three years ago, people thought they could do massive replacements of technology, such as moving everything to SaaS applications. “Now we’re seeing way more adding of technology rather than replacing,” he says.
The result can be a tangle of overlapping, redundant systems that cost money, slow innovation and hinder new business opportunities.
“I’ve been in discussions with CIOs who want to replace something, but they can’t be sure of what would happen if they did,” Burger says. But “sometimes you just have to turn something off and see who yells”.
We spoke with five CIOs who did just that and lived to tell the tale.
Understanding undocumented knowledge
Sometimes there are demands from the business to move quickly, but not necessarily practically or prudently. Sometimes there are systems left over from mergers, acquisitions and internal reorganisations that haven’t been sunsetted. Sometimes systems multiply as CIOs try to keep the lights on and innovate simultaneously.
“Complexity is the result of a diversity of footprints, of tools, of workforce,” says Christopher Rence, CIO of Digital River, a US provider of e-commerce, payments and marketing services for merchants. Rence knows whereof he speaks: He’s lived through three acquisitions in four years and seen the residue of the 20 acquisitions experienced since 1994.
“One company we acquired had nothing but white-label hardware. It didn't have an asset value, but it was doing a lot of processing,” he recalls.
In preparation for conducting a strategic migration of the data through a gateway into a SaaS solution, “we had to do a full inventory of what those homegrown products were doing,” says Rence.Read more: Gartner to enterprises: ‘May the algorithmic force be with you’
“It required understanding some of the undocumented knowledge. We had to sit down with the programmer who originally wrote the integration scripts and ask him to create industry-standard processes and APIs so they could be reusable and repeatable.”
2. Taking a consolidated cloud approach
The Weather Company’s Koehler traces complexity’s origins to illogical architectures -- multiple systems performing similar functions, too many layers of a stack serving extraneous purposes. These made operations costly, risky and difficult to change.
The Weather Company had 13 interdependent data centres (11 in the US, one in Asia, and one in Europe). “If one application needed data in another data centre to work, we weren’t lowering risk. We were increasing it by spreading out our dependencies,” Koehler explains.
“There were also four forecasting systems, each one producing a different result.” His team whittled that down to one forecasting system, a common API platform for sending data out to partners, and a consolidated cloud-based approach for infrastructure.
“Moving our forecasting system to Amazon’s infrastructure saved us seven figures over the lifespan of the technology investment that we would have had to make,” says Koehler. Even better, it made his group more agile. “We inverted the ratio of time spent on maintenance from 80 per cent to 30 per cent. You have to get yourself in shape before you can innovate.”
That agility brought other value, he notes. His company's widget became the default source for weather on all Apple devices running iOS 8.
3. Building inventory system tools
Atish Banerjea, CIO of NBCUniversal, encountered a high level of complexity that developed under previous ownership when divisions deployed a lot of shadow IT. When he arrived at NBCUniversal in December 2012, his team partnered with ad sales to consolidate 19 different advertising selling applications into one.
The result is a much higher level of efficiency. “We’ve built tools so we can see inventory across units. We have a much more accurate inventory of what we can sell,” he said. “We can do forecasting faster, and it’s helped us improve pricing.”
Similarly, each division had its own Web hosting provider. NBCUniversal’s technology team led an effort to consolidate all 200 Web properties onto one platform. “Across all these initiatives, we have saved tens of millions of dollars,” Banerjea says.
4. Removing excess points of integration
Consolidating a panoply of old systems not only gives you technology that’s easier to audit, it also makes it easier to roll out new systems. “When you have so many points of integration and dependency, it makes it that much more difficult to find out how and where errors occur,” says Henry Jenkins, director of information services at Huntington Hospital in California.
“By reducing complexity, you have fewer mistakes and more repeatable outcomes with little variance. When something does go wrong, less complexity makes it easier to figure out what went wrong, correct the root cause, and move on.”
In the healthcare industry, the conversion to electronic medical records is a challenge. “But we have an opportunity to have a more complete medical record that follows the patient,” says Jenkins. “Instead of rooting around in twenty different departments and 100 different spreadsheets, we can improve outcomes by making better decisions with more accurate information.”
5. Adding safeguards against seepage
Unless there is a proper function to manage it, usually done through enterprise architects, to ensure there’s an appropriate swap of applications -- that is, an old application is actually replaced, not augmented -- complexity will spread like mould in an old bathroom, spreading behind the walls and under the floors, unseen and insidious. That’s why CIOs who do the work of simplification must put in place safeguards against that seepage.
That’s why Banerjea brings his enterprise architecture team into every budget meeting with a “living, breathing framework of our inventory across all our units”. If one division proposes something new, the EA team is there to identify commonalities. With visibility across the company’s tech stacks, “we can see when we can leverage an application for multiple uses and avoid starting from scratch”.
The operative word in all this is courage. “You don’t do it tepidly. You have to have a full commitment to making the change, because it will help you move to a fundamentally new place,” Burger says. “Change management isn’t tied to an event anymore. It has to be a core ability for CIOs.”
Koehler agrees. “It boils down to having courage to call somebody’s baby ugly, turn it off and force good decisions throughout the life of that new system.”