娇色导航

Nothing in the world is perfect, and that includes software code, too.

In fact, most programmers aren’t given the time to create even near-perfect code. With the pressure to deploy fast, they often release bit of problematic code knowing that it will have to be fixed in the future.

That creates technical debt.

“We see a ton of tech debt from a code perspective. [Programmers] are erring on the side of speed and innovation, and they’re implementing code that isn’t QA’ed as well as it should be. That is creating issues down the road,” says Nate Buniva, a partner at digital services firm West Monroe. “And that’s happening more and more.”

Although the term “tech debt” is sometimes applied to any and all outdated IT and is often seen as synonymous to “legacy tech,” tech debt originally referred to code debt specifically and in many tech circles is still seen as such.

However, just as CIOs must clear out the legacy components in their IT stack that create drag on their business strategies, pose risk to their security, and add costs to their budgets, CIOs similarly must contend with technical debt.

The costs for not doing so can be high. The most recent study available, from 2022, put the cost of poor software quality at $2.41 trillion in the United States alone.

Given that reality, there’s a push for CIOs to implement practices to keep their tech debt in check. Here, experienced technology leaders share six strategies they use to trim their technical debt.

1. Get analytical about measuring your technical debt

, research director for the infrastructure and operations practice at Info-Tech Research Group, is a strong proponent of cataloging technical debt. The analyst advises IT leaders to establish a list of their critical technical debt, know the business impact of that debt, and have a process for addressing it.

Many CIOs, he and others say, too often fall short on these three foundational issues.

“One of the biggest challenges is just understanding and organizing the scope of technical debt,” Sharp says, explaining how IT teams struggle with knowing the amount and impact of technical debt “because it sprawls into different systems. It has knock-on effects.”

But like most everything else in business today, debt can’t successfully be managed if it’s not measured, Sharp says, adding that IT needs to get better at identifying, tracking, and measuring tech debt.

“IT always has a sense of where the problems are, which closets have skeletons in them, but there’s often not a formal analysis,” he says. “I think a structured approach to looking at this could be an opportunity to think about things that weren’t considered previously. So it’s not just knowing we have problems but knowing what the issues are and understanding the impact. Visibility is really key.”

Sharp cautions against tracking every bit of tech debt, though, stressing instead the need to track the debt intended to be fixed.

2. Don’t give AI-generated code a pass

Workers in IT and, indeed, throughout the enterprise are using generative AI to write code. Some research has found that . However, most organizations can expect to “see a glut of very mediocre software over the next year or two” instead, says , chief AI officer at Thoughtworks, which provides software design and delivery as well as consulting services.

“AI lets you crank out code rapidly, but [people] seem more adapt at cranking out new code rather than fixing it,” he says, noting that too much of that mediocre code is being moved into production without review by expert developers or automated quality assurance checks.

“We predict organizations will get themselves into trouble, where they have more features get through faster, but it will be code bloat and then more tech debt and legacy code,” Mason adds.

That doesn’t mean workers shouldn’t be using AI for coding, Mason says. Rather, organizations should ensure they have the processes and tools — including AI-enabled ones — to ensure they’re not letting an unacceptable amount of poor-quality code slip through.

3. Apply governance to tech debt

Most organizations have some governance around their software development programs, Buniva says. But a good number of those governance programs are not as strong as they should be nor detailed enough to inform how teams should balance speed with quality — a fact that becomes more obvious with the increasing speed of AI-enabled code production.

“Governance is not keeping pace with gen AI,” Buniva says. “You need governance that is appropriate for gen AI, one that doesn’t slow innovation but doesn’t allow for creating a ton of tech debt.”

A good governance program should set requirements for testing and QA as well as specify when humans should be involved instead of automated QA decisions, Buniva says. It should also address training requirements so that anyone developing code is educated on the standards.

4. Prioritize what you pay down

Like legacy tech more broadly, code debt is a fact of life and, as such, will never be completely paid down.

So instead of trying to get the balance to zero, IT exec Rishi Kaushal prioritizes fixing the most problematic pieces — the ones that could cost his company the most.

“You don’t want to want to focus on fixing technical debt that takes a long time and a lot of money to fix but doesn’t bring any value in fixing,” says Kaushal, 娇色导航of Entrust, which makes identity and security solutions.

He focuses on addressing tech debt that poses security risks or that creates friction in users’ experiences — the same approach he takes to legacy tech as a whole.

“Some tech debt is OK to have; it’s OK to keep. So you have to decide what is the tech debt you can live with while you make other things better,” he adds.

, global head of technology, media, and telecommunications at professional services firm KPMG, offers a similar strategy, advising “super targeted, proactive investments, and quick course corrections” when dealing with any legacy tech, including code debt.

He says most CIOs know where those pain points are. He points to findings in his firm’s 2024 Global Tech Report showing that 57% of responding organizational leaders say that flaws in their foundational IT systems disrupt business-as-usual on a weekly basis.

Fixing those problematic systems is a good place, Gibson says, adding that CIOs could use logs and surveys of their IT staffers to identify other problems that should top the to-fix priority list.

And when it comes to gaining backing for tackling tech debt, IT leaders should sell the C-suite on tech debt being a business risk.

5. Get specific when setting goals

, senior vice president and 娇色导航for North America at Schneider Electric, has to cut legacy tech by 12% by end of 2025. “We have clearly defined target states,” he says.

Having a specific target gives Cain and his team a glaring incentive to modernize systems and address any problematic code in the process.

To reach that goal, Cain has devised a multipronged strategy that includes cataloging and measuring his current state so he knows where that legacy tech/tech debt resides as well as prioritizing and mapping out action.

Although such work is foundational to any well-run IT function, Cain acknowledges that having an assigned objective keeps the fire lit under the team. As he notes, “You can’t drive change without a forcing function.”

6. Recognize debt management is an ongoing process

Wayne F. McGurk, 娇色导航and SVP of IT for the National Rural Electric Cooperative Association, doesn’t see technical debt as a good or bad thing but rather “a natural outcome of the development process, occurring because something new is being built.”

“There’s a tendency to go as fast as you can to get the MVP [minimum viable product] out there, and you don’t necessarily build an overly industrialized application at the beginning,” he says. Teams make tradeoffs, opting for technologies that work for the MVP that they know will be insufficient as solutions scale.

So McGurk factors that into not just his development cycle but IT operations, pulling in various tactics to create a holistic approach for managing technical debt on a continuous basis. As part of this approach, McGurk’s team documents and details the introduction of any new technical debt, which is then tracked through the organization’s ticketing system so that IT teams “can pull that all up and report it and look at it.”

McGurk also considers how each piece of technical debt impacts operations in five key areas: simplicity, flexibility, continuity, security, and transparency.

“When technical debt starts hindering any of those operating principles, then it’s risen to the level where we want to address it,” he explains.

McGurk and his IT team consider the level of impact, the risk to the organization, and the organization’s overall strategy to then prioritize what needs attention. They then make those determinations known, thereby creating visibility into the topic across the organization.

All this gets wrapped into his IT department’s workflow, McGurk says, which ensures managing technical debt isn’t treated as a one-off project but is instead managed in an ongoing manner. For example, his Scrum teams are expected to identify new technical debt and determine when and how to address it.

“You have to build the culture of accountability and responsibility so your teams know that just because a project is delivered, it’s not done. It’s a journey, and there’s no end to it, so then it becomes part of your demand management strategy — handling both the demand for new work but also legacy work and technical debt,” he says.