top of page

3%: What Gartner's Board Number Really Says About Work Design

Oct 29
5 min read

Boards expect AI to cut costs. Gartner says almost no one gets there through productivity. Its own analyst explains why, and the explanation is about workflows, not technology.

This week Gartner put a number on the gap between what boards expect from AI and what organizations actually get. 77% of board members expect AI to deliver productivity and cost savings. Just 3% of organizations exceed their cost-savings goals through productivity.

Most coverage will read that as an AI story. It isn't. Read past the headline and Gartner's own diagnosis points somewhere else: the capacity AI creates is real, and organizations have no way to put it to work. That is a work design problem. And with budget season about to start, it is the most useful number a practitioner can bring into the room.

What Gartner actually found

The figures came out of the opening keynote at Gartner's HR Symposium/Xpo in Orlando on October 26, 2026.

Finding

Figure

Board members who expect AI to deliver productivity and cost savings

77%

Organizations that exceed cost-savings goals through productivity

3%

Companies already over budget on AI

38%

Companies planning to spend more on AI next year

85%

CHROs reporting no leadership alignment on where AI creates value (July 2026 survey, n=297)

78%

Share of employee time AI is freeing up today

About 1 in 5

A note on the headline number: some outlets, including The Register, report it as 75%. Gartner's own release leads with 77% and describes it as "three-quarters" in the body. 77% is the figure to cite.

The pattern is hard to miss. Spending is up, budgets are blown, and the savings aren't showing. Yet nearly everyone plans to spend more.

Read the diagnosis, not the headline

The most important line in the coverage isn't a statistic. It's how Gartner senior director analyst CV Viverito described the problem. Organizations, he said, are not struggling to create capacity. That is already happening as they redesign work. What they are struggling with, per The Register, is "how to make that capacity usable."

Sit with that. The technology is doing its part. Hours are being freed. The failure happens after the hours exist, in the space between "someone has more time" and "the organization gets more value."

That space has a name. It's the workflow. Gartner's own keynote highlights make the same point from the other direction: the organizations seeing the strongest results are using redesign to change the cost of work, and they are more than twice as likely to exceed their cost goals. The tool didn't separate winners from losers. The redesign did.

Why freed capacity evaporates

If AI saves an analyst five hours a week, where do those hours go? In most organizations, nowhere in particular. Three things happen.

  1. The minutes are scattered. AI saves twenty minutes here and forty there, spread across hundreds of people. Fragmented time doesn't add up to a role you can stop funding or a project you can now staff. It gets absorbed by the next meeting, the next thread, the next request.

  2. The bottleneck is somewhere else. Speeding up a step that wasn't the constraint doesn't speed up the workflow. If a deliverable still waits three days for an approval, drafting it in an hour instead of four changes nothing the customer or the P&L can see. The work just waits longer in the queue.

  3. There is no destination. Gartner found that 78% of leadership teams don't agree on where AI creates value. Without a shared answer to that question, there's nowhere to point recovered capacity. People fill it with whatever is loudest.

None of these are AI failures. They are coordination failures that existed before AI arrived. AI just made them expensive enough to show up in a board survey. This is workflow debt coming due.

Productivity is the wrong unit

The board expectation behind the 77% assumes a straight line: AI makes people faster, faster people cost less, costs go down. The 3% says the line breaks somewhere in the middle.

The Work Management Institute's Work Value Pyramid explains where. It separates three levels of value: Activities (work done), Progress (work moving toward a result), and Outcomes (results the organization actually needs). AI is very good at the bottom layer. It makes activities cheaper and faster. But cost savings live at the top. They come from outcomes delivered with fewer steps, fewer handoffs, and less rework.

Activities only climb the pyramid through a workflow. If the workflow isn't designed to absorb faster activities, the gains stay at the bottom. Hours get saved, and the organization is no better off.

That is why "how much time did AI save?" is the wrong question for budget season. The right one is: "which workflows changed shape because of it?"

What making capacity usable looks like

Gartner's recommendations point the same way: concentrate AI on the work that matters most, spread AI know-how across the business, and give managers room to redesign the work around the technology. In workflow terms, that means four things.

  • Decide where capacity goes before you create it. Name the outcome a team's freed time should serve. The Coordination Stack's first question, Why, belongs at the start of every AI rollout, not after it.

  • Redesign the workflow, not the task. Savings show up when a step, a handoff, or an approval loop disappears. Map the end-to-end flow, find the constraint, and aim AI there.

  • Measure flow, not activity. Cycle time, wait time, queue size, and rework tell you whether the work got faster. Hours saved per person tells you almost nothing. These are the Workflow Performance Indicators (WPIs) WMI uses, and each needs a named owner.

  • Treat redesign as ongoing. Gartner's own July 2026 research argues that AI value stalls when organizations treat work redesign as a one-time change. Workflows drift. Someone has to own them.

This is the core of work management as a discipline: making work clear, coordinated, and complete. AI raises the stakes on all three.

Questions to bring to budget season

When someone asks "why aren't we seeing savings from AI?", the 3% gives you cover to change the conversation. Instead of defending the tools, ask:

  1. Which three workflows matter most to our results, and has AI changed how any of them run end to end?

  2. Where does work wait longest in those workflows? Is AI pointed at that constraint, or somewhere easier?

  3. When AI frees up time on a team, who decides what it goes to?

  4. Do our leaders agree on where AI should create value? If not, what forum will settle it?

  5. Are we measuring cycle time and rework, or just adoption and hours saved?

  6. Who owns each redesigned workflow after launch?

If the answers are vague, more AI spend won't fix it. 85% of companies are about to find that out the expensive way.

The real signal

The 3% isn't proof that AI doesn't work. It's proof that AI can't fix unclear priorities, can't coordinate across teams, and can't compensate for broken workflows. Gartner's analyst said it plainly: the capacity is there. Making it usable is a management job.

The most important skill of the AI age will not be using AI. It will be managing work.

Sources

 
 
 

Comments


bottom of page