Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Transformational AI ROI requires a work redesign

The path to real returns runs through your constraints.

[Photo: Getty Images]

Profile photo of Andy MacMillan

We’re now at the point where most companies use AI, but it’s difficult to tell whether they’re using it effectively. When I speak with other executives, they usually have good things to say about AI. Employees work faster, reports run quicker, and information is easier to find.

The problem is that no one can connect these wins to numbers on a balance sheet. In my conversations, I’ve observed three tiers of AI ROI:

  • Soft ROI: Employees are more effective in general, but there’s no direct impact on metrics.
  • Hard ROI: AI-related gains are measurable, such as hours saved or tasks automated.
  • Transformational ROI: An organization fundamentally redesigns itself to be more efficient and profitable.

Almost every leader I’ve spoken to wants the third outcome but is currently stuck somewhere between the first two. My advice is to see AI ROI as a constraints problem, not a technology problem.

The theory of constraints

The Theory of Constraints first appeared in Eliyahu M. Goldratt’s book The Goal. According to Goldratt, improving business efficiency relies on finding and addressing bottlenecks in a process. If you spend time and resources on improving parts of the system that aren’t actually constrained, you won’t improve your overall performance that much.

Suppose your company’s smartphone app lets customers manage their own accounts. Your software engineers’ new AI tool cuts coding time for app updates in half. In theory, AI has delivered enormous ROI by freeing up staff for other high-value assignments.

In practice, though, every patch still needs to go through product management, approvals, compliance, and deployment. You’ve spent thousands of dollars (or more) to make one part of the pipeline faster, but coding time was never your constraint. Instead of fixing a bottleneck, you just created a backup of more work stuck at the bottleneck.

So many AI conversations center entirely on time savings or headcount reduction, which I tend to push back on. Even if people are working more efficiently, I wonder whether they’ve removed any real constraints.

A 4-step transformational ROI framework

When it comes to AI, soft ROI means you’re off to a good start, and hard ROI means that you can justify your expenses. For transformational ROI, you need a bigger framework that identifies and removes your bottlenecks.

1. Start with the balance sheet

The C-suite defines the business outcome it wants to achieve, whether that’s revenue growth, margin expansion, faster product launches, staff retention, or operational efficiency. Focus on one goal at a time so you can see how changing one system affects the others.

2. Find the constraint

Executives should partner with domain experts to figure out why the business can’t achieve the desired outcome right now. Look for places where work slows down, decisions stall, and information gets muddled. Intelligent enterprises can leverage trustworthy, transparent data to find bottlenecks.

3. Redesign the work

Consider what the workflow would look like if AI had been available from the start. Move beyond automation and imagine a whole new operating model. This is the most difficult part of the process, but it’s also where you and your staff get to be creative.

4. Deploy AI against the new operating model

Once you’ve identified bottlenecks and redesigned your workflows, you’re ready to field-test your AI solutions. Agents, analytics, and automation are all useful technologies to try. Remember that your goal is to build a better process, not just save time.

While executives can kickstart this process, they’re probably a few steps removed from the workflows that need revamping. To reimagine those workflows completely, talk to the domain experts who use them every day.

Domain experts offer the key to success

The domain experts at your company may or may not be AI pros (yet), but they know exactly what works and what doesn’t in their day-to-day assignments. Getting to transformational ROI means giving these experts room to experiment and permission to fail.

Encourage them to use their imaginations. If a critical workflow disappeared tomorrow, ask how they would redesign the process from scratch. Many of their ideas won’t work. The ones that do could eliminate constraints they didn’t even realize were there.

As a thought experiment, imagine a regional bank with a home mortgage division. The system’s bottlenecks include everything from calling new applicants to gathering signatures for real estate contracts. AI could speed up individual steps, but the overall process would still be slow and tedious.

A redesigned, AI-native workflow might include new customers submitting their information via online forms, and AI agents immediately initiating background checks and researching credit histories as permitted by regulatory guidelines.

Sophisticated chatbots could lay out the basics for approved customers, even after hours or on a weekend. After that, mortgage brokers might still want to talk with customers directly, but those customers would be well informed, and the conversations could be much shorter.

When a customer finally settles on a place to buy, AI email tools can send out all the required paperwork, gather e-signatures, and ping the broker when everything is done. If online signatures aren’t practical, agentic AI could also coordinate everyone’s schedules and find a convenient time to meet in person.

The shift from “a broker does most of the administrative work” to “AI agents do most of the administrative work” represents a new operating model. Your company’s domain experts can think of unconventional ideas like these, provided they have C-suite support.

Ask AI better questions

Getting transformational AI ROI isn’t really a question of technology. It’s a question of whether an organization is willing to identify bottlenecks and remove them, even if it means redesigning comfortable workflows. The companies that achieve transformational ROI with AI will pair executive clarity on outcomes with domain expertise in systems.

Stop asking AI to do old work faster. Start asking it to make new work possible.

Andy MacMillan is CEO of Alteryx.