7 ways to kill IT’s value to the business

Most CIOs have got the message that their job is largely about delivering business value, and less about system uptime and other operational metrics. New transformational opportunities emerge every two years, and the question is whether IT is consistently delivering value.

Agile, DevOps, and ITSM practices aim to improve delivery and service capabilities. Each successful deployment raises the bar for stakeholder expectations. Delivering value consistency can be elusive, and common issues include managing employee turnover, addressing technology partners who miss expectations, balancing security priorities, and addressing technical debt.

CIOs are now under fire to deliver business value from gen AI. The C-suite wants more than experiments, and expects short-term ROI and longer-term strategic value from AI investments.

I’ve recently written about rethinking the IT organization in the gen AI era and what world-class IT looks like. CIOs must also look out for the practices and behaviors that kill IT’s value to the business. Experts weighed in on six of them.

1. Targeting big-bang deployments

AI’s hype has raised executive expectations around delivering gen AI capabilities. I’ve heard executives targeting autonomous operations, advanced agentic AI capabilities, and personalized, AI-enabled customer journeys.

Underperforming IT departments can fall prey to demanding stakeholders. Product-based IT departments don’t commit to big-bang deployments and have the discipline to experiment, prioritize minimally viable products (MVPs), and embrace agile delivery of capabilities that improve incrementally.

 “When planning and executing digital transformation initiatives, brands should start small and scale without disruption,” says Raj Balasundaram, global vice president of AI innovations for customers at Verint. “The primary goal should involve realizing measurable outcomes quickly rather than waiting for a multi-year project.”

Recommendation: IT departments will find it easier to partner with their stakeholders on agile delivery by inviting them to sprint reviews and scheduling frequent brainstorming sessions.

2. Championing AI POCs without a deployment plan

Top-tier SaaS and security companies are releasing AI agents to uplift workflows. CIOs need the experiments to understand which capabilities deliver value, where data quality improvements are needed, and how to scale a rollout.

Unfortunately, many IT departments are not working directly with end users on these experiments. Others undercommunicate an agile plan for what’s required for a full deployment. The result can be missing stakeholder expectations or having many POCs that fail to reach production.  

“Too often, teams launch a flashy pilot in a few weeks and accidentally teach the business that AI is simple and instant,” says Stéphan Donzé, founder and CEO of AODocs. “Then everything slows down as they confront reality with no governed data pipelines, no plan to keep knowledge current, no controls to prevent leaks of confidential information, and no strategy for managing model costs.”

Pilots need clear objectives, and teams should consider when to pivot or end floundering experiments. When reaching a successful milestone, IT should have an approach to scale pilots to more data, users, and use cases.

Kurt Muehmel, head of AI strategy at Dataiku, adds that it’s a mistake for IT teams to treat successful pilots as proof their agents will work in production. “Pilots succeed because they operate without real constraints, such as small datasets, forgiving users, and human oversight when things break.”

Recommendation: IT should co-author a vision statement to define the success criteria for any POC, then apply agile to manage the POC and multiple pilot phases. These disciplines ensure stakeholders are part of the journey and convey that experimentation is one of the deployment milestones.

3. Focusing on deployment and not adoption

Achieving buy-in for iterative deployments is necessary to deliver consistent business value, but it’s not sufficient. Too many DevOps and data science teams consider it a “job well done” when deployments are on schedule and achieve the targeted scope. But this fails to recognize IT’s change management responsibilities, especially when AI capabilities can dramatically reshape how departments operate.

“IT teams are highly skilled, but they’re often not the end users of the products they build, and they rarely get enough feedback from those who are,” says Nik Froehlich, founder and CEO of Saritasa. “The result can be perfectly functional solutions that no one actually uses. The fix is simple: Involve end users early and often, listen to their feedback at every stage, and build solutions that truly work for the people using them.

Recommendation: One reason CIOs should review their agile practices in the AI era is to redefine the definition of ‘done’ so that agile teams include adoption and change management as feature-level acceptance criteria.

4. Prescribing the future of work

Boards expect CIOs to communicate an AI strategic governance model, including a charter outlining how employees should experiment with AI capabilities. Governance should define which AI tools employees can use, which data they can access, and where they should report on the successes, learnings, and failures of their experiments.

CIOs should assign architects, business relationship managers, and analysts to collaborate on these experiments. But stepping over the grey line and prescribing departmental workflows can be a value killer, as IT is rarely versed in the operational goals and workflow exceptions.

“Real change is now bottom-up, driven by how people actually work, not by how the organization prescribes they should work,” says Brian Madden, VP, technology officer, and futurist at Citrix. “The top-down, multi-year roadmap model broke the moment workers began adopting AI tools on their own, yet most companies today still don’t provide a safe, governed way for workers to use third-party AI. When there’s no sanctioned path, people use these tools anyway, and unmanaged AI adoption increases risk without delivering strategic value.”

Recommendation: CIOs must communicate a top-down strategy, but the No. 1 reason digital transformations fail is when IT doesn’t recognize that evolving operations require bottom-up participation. IT departments should assign Six-Sigma-trained process experts to partner on experiments and serve as guides in transforming workflows with AI agents and people collaborating.

5. Accelerating technical debt when deploying AI

SaaS sprawl is an issue facing many IT departments, especially when driven by department leaders who select their own tools and foster shadow IT practices. While organizations should be experimenting with AI, deploying rogue AI agents is a security and compliance risk.

Weldon Dodd, distinguished engineer at Iru, says, “Companies today are buying various AI tools to solve specific problems, and over the next few years, this proliferation will result in IT teams having way too many vendors to manage.”

IT architects should communicate standards and promote the use of extendable platforms as key strategies. While there will always be pressure from stakeholders who want to prescribe technologies and tools, architects must communicate tradeoffs, the costs of supporting new platforms, and the risks of accumulating technical debt.

“The IT team curtails its potential impact when its AI and data initiatives are built in a vacuum without asset visibility and a shared view of how systems, such as gen AI, both create value and introduce exposure,” says Yakir Golan, CEO of Kovrr. “Effective governance thus hinges on understanding what tools exist, what data they interact with, and the implications of their failure.”

Recommendation: While business teams often try to prescribe solutions, it’s IT’s responsibility to ask questions and discover the targeted business needs and requirements. CIOs should help IT leaders develop the business acumen, the courage, and the patience to collaborate with stakeholders on goals, rather than accepting solution demands as marching orders.

6. Accepting one-time projects without ongoing support

When agile teams successfully navigate AI from ideas through POCs, production deployments, and end-user adoptions, then they are really just at the start of delivering business value. The next stages require capturing feedback, measuring results, and iteratively making improvements to deliver against the stated success criteria.

Agile teams want to iterate, and stakeholders generally want to see improvements in technology, data, and AI capabilities. So why do IT departments have so many legacy applications, accumulate technical debt, and have frustrated end users seeking application modernization?

“Many IT teams don’t plan beyond implementation and fund continuous optimization, including ongoing testing, user adoption, data governance, and iterative improvement that can turn even capable systems into shelfware,” says Pankaj Goel, CEO and co-founder of Opkey. “The strongest transformations succeed when IT shifts from ‘deploying technology’ to building a dynamic, optimized transformation that evolves with the business.”

Recommendation: A key responsibility of the agile PMO is to enforce financial governance with departmental leaders and the CFO. Initiatives that lead to successful deployments and delivering business require ongoing funding for upgrades and other lifecycle management needs.

7. Underinvesting in time for learning

Many CIOs have a budget for assigning courses, attending conferences, and coaching leaders. But subscribing to learning programs isn’t sufficient, as many IT employees are under pressure and time constraints to commit to training. Additionally, employees need time to develop lifelong learning practices by experimenting, experiencing, and teaching new skills.

“High-performing teams include learning in their long-term plans so skill growth and innovation happen together,” says Michael Pytel, senior technologist at VASS. “Challenge your technical team to spend one hour a week learning, offer reimbursements for the testing fee typically required for certifications, and incentivize lunch and learn sessions where team members share what they learned with those around them.”

Recommendation: CIOs should reflect on ways IT can kill business value, as well as ways leadership can kill IT culture. Leaders can’t push IT to run at 130% capacity indefinitely; they must create learning opportunities and periods to enable process evolution.

CIOs reading this article should recognize that many of the issues I identified boil down to instituting and adopting strong governance and a flexible operating model. Additionally, CIOs must coach their leaders on partnering with stakeholders to deliver business value while enforcing non-negotiables across governance, operating principles, and collaboration responsibilities.