By Cheryl Howard, MSPH, PMP, PMI-CPMAI, Prosci Certified Change Practitioner
October 15, 2026
The AI demo works. The AI team is excited. Someone reports that a task now takes 40 minutes instead of four hours. Then someone asks the question: “Where does this show up in the P&L?”, and the room goes quiet.
That silence is common. MIT’s Project NANDA found that about 95 percent of the enterprise generative AI pilots it studied produced no measurable P&L impact, and its authors pointed to integration and learning gaps rather than model quality[1]. McKinsey’s 2026 State of AI survey tells a similar story from another angle. Only 37 percent of respondents attribute any Earnings Before Interest and Taxes (EBIT) impact to AI, unchanged from the year before[2].
I read these numbers as a signal that we have not been managing AI for benefits. Project, program, and portfolio professionals already own the discipline that closes this gap. We just have not applied it to AI with the rigor we would bring to an ERP rollout or a new manufacturing line.
Time Saved Is Not a Benefit Yet
Here is the trap. A pilot measures output: hours saved, documents drafted, tickets closed. The P&L measures outcomes, such as cost removal or revenue gained. Between the two sits a decision someone must make.
If an analyst saves six hours a week, does the organization absorb more volume without hiring, redeploy that time to higher-value work, or end a contractor engagement? If nobody decides, the six hours disappear into the week. The pilot looks successful. The P&L never moves.
PMI describes benefits realization management as identifying benefits, aligning them with strategy, and ensuring they are realized and sustained after implementation is complete.[3] That last clause matters most for AI. Pilots end. Benefits are earned in operations.
Managing AI for Benefits
Managing AI for benefits requires more than tracking efficiency gains. It means defining measurable business outcomes, assigning clear accountability, redesigning workflows, and continuing to verify that benefits are realized and sustained over time. Let us break that down.
Name a benefit owner before the pilot starts. Not the data science lead or the vendor. It should be the person in the business whose budget or KPI will change. In regulated environments, I would also want the quality and internal compliance owner named, because a benefit that creates inspection risk is not a benefit.
Write the benefit hypothesis in financial language. “Reduce deviation investigation cycle time by 20 percent, freeing two FTEs of QA capacity to support the new manufacturing line without incremental hiring” is something a CFO can track. “Improve efficiency” is not.
Baseline before you build. Teams try to demonstrate improvement against a starting point that may not have been measured. If you cannot baseline today’s cycle time, error rate, or cost per transaction, you cannot credibly claim any of them improved.
Redesign the workflow, not just the task. McKinsey found that workflow redesign had the biggest effect on EBIT impact from generative AI among the attributes it tested. They found tracking well-defined KPIs was the adoption practice with the most bottom-line impact[4]. The takeaway: AI delivers greater business value when organizations rethink how work gets done and measures the results clearly defined KPIs.
Match the benefit to accountability. This is where I apply the Governed AI Test: delegate, review, or lead. Delegated automation can usually be measured on throughput and cost. Assisted work must net out the review time it adds. If a reviewer spends 20 minutes checking a draft that saves 30 minutes, the real gain is 10 minutes. Augmented work, where AI shapes judgment, is often measured in decision quality and risk avoided. These benefits take longer to surface.
Keep the benefits register open after go-live. PMI’s framework calls for verifying that benefits have been delivered and monitoring actual results against targets, once the capability transitions to the business.3 A benefits register is a structured list of all planned benefits, including their owners, metrics, baselines, targets, and dependencies. For AI, that register needs a monitoring trigger. A well-maintained benefits register is essential for ensuring long-term benefits sustainability.
The Leadership Conversation
When I led modernization programs across the drug development lifecycle, the programs that held their value had one thing in common. Someone in operations owned the result long after the project team moved on. AI is no different. It simply moves faster, so the window between a promising pilot and an abandoned one is shorter.
Before approving the next pilot, I would ask three questions: Who owns the benefit? What will we stop doing, spend less on, or do more because this works? How will we know in twelve months, not six weeks?
If the team can answer, you have a pilot that is scalable. If they cannot, you have a demo.
Pilots prove possibilities. Benefits realization proves value. Our profession already knows how to do the second part. It is time we insisted on it for AI.
References
[1] Challapally, A., Pease, C., Raskar, R., & Chari, P. The GenAI Divide: State of AI in Business 2025. MIT NANDA, MIT Media Lab (July 2025). Available by request from MIT Media Lab: https://www.media.mit.edu/groups/nanda/overview/
[2] McKinsey & Company. The State of AI in 2026: On the Road to ROI (2026). https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
[3] McKinsey & Company. The State of AI in 2026: On the Road to ROI (2026). https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai
[4] McKinsey & Company. The State of AI: How Organizations Are Rewiring to Capture Value (2025). https://www.mckinsey.com/~/media/mckinsey/business%20functions/quantumblack/our%20insights/the%20state%20of%20ai/2025/the-state-of-ai-how-organizations-are-rewiring-to-capture-value_final.pdf
Cheryl Howard
Cheryl Howard is the Founder and CEO of Howard Consulting LLC, a boutique consulting firm specializing in change management and digital transformation for life sciences, healthcare, and high-growth industries. She leverages her PMP® and Prosci® Change Management certifications, designing and implementing tailored strategies in complex regulatory environments.
Cheryl brings a practical, people-centered approach to every engagement. She believes people are at the heart of every transformation. In her firm, she builds trust, listens actively, and designs solutions that work – not just technically, but culturally. Howard Consulting LLC’s mission is to help organizations make better decisions, work more efficiently, try new things, and feel prepared for the ever-changing future.
Before launching Howard Consulting, Cheryl was the Director of IT Strategy at Gilead Sciences, where she played a pivotal role in optimizing the company’s drug development lifecycle -spanning discovery, manufacturing, and commercialization – supporting over $30 billion in annual revenue.
Connect on LinkedIn: https://www.linkedin.com/in/cheryl-howard25/
