Field notes for IT, PMO and transformation leaders

Deciding what to fund, and knowing what it returned

Over the past few months I have spoken with IT, PMO, data and transformation leaders about how technology investments are chosen, funded and measured afterwards. Five patterns came up again and again. Next to each I have placed what recent research says, so you can test what I heard against a wider base.

14%
of organizations fully captured the value targeted by their highest-priority cost initiative last year.
44%
of companies fund the next AI wave from earlier automation savings that came in below target.
24%
of organizations account for value leakage when they set savings targets.
31%
of CIOs have rules for stopping AI pilots that are not delivering.
1  Where attention goes

The key programs get the scrutiny. The rest of the portfolio is approved and left alone.

What I hear

Large, visible programs get a business case, steering reviews and benefit tracking. The long tail is approved once and rarely revisited, even when hundreds of millions are committed each year.

Cases from different teams are built differently, so comparing them at the funding table takes judgment more than evidence.

What the research shows

BCG's 2026 study of nearly 600 multi-business companies finds that the best capital allocators weigh investment requests against one another, rather than approving each on its own merits. Comparing cases side by side turns a string of go or no-go calls into a real selection. BCG, Aug 2026

In BCG's survey of 152 CEOs, only 14% have clearly defined the P&L impact for all of their AI initiatives. BCG's advice: define the value logic before each initiative launches, with finance validating results from day one. BCG, Jul 2026

What teams can do

Apply one lighter, consistent standard to every case, not only the big ones, so the long tail can be compared at approval and revisited later.

2  What the returns look like

Most returns land as productivity, which is the hardest benefit to put on a P&L.

What I hear

When leaders are candid, most of the benefit in their cases is productivity: hours saved, faster cycles, fewer handoffs. Finance discounts it, because hours saved do not show up until someone decides what to do with them.

Costs are often understated too. In one case shared with me, software pitched at $150K a year needed about $1.7M a year more in support and infrastructure that never appeared in the case.

What the research shows
Lean toward productivity45%
Lean toward decision quality20%

Gartner found finance AI investments lean heavily toward productivity, while boards put more weight on growth and better decisions. Gartner, Jul 2026

Across functions, productivity is the most common AI objective and takes about 30% of functional AI spend, nearly twice the next objective. Gartner, Sep 2026

Half of executives say AI spending is approved without adequate visibility into its full cost, and 59% saw AI costs run significantly over expectations. Capgemini, Nov 2026

What teams can do

State up front how a productivity gain will be captured (released capacity, avoided hires or reduced external spend) and cost the full lifecycle, so finance can accept both sides of the case.

3  How to sequence AI investment

Leaders split between starting broad and small, and concentrating on a few big bets.

What I hear

One CIO told me he went with small productivity gains across a broad base, such as Copilot, to build confidence before moving to larger bets. Others are going straight to a few large programs.

Start broadMany small gains, low risk, quick to show. Harder to prove at the P&L.
ConcentrateFew high-value areas, multi-year funding. Bigger payoff, bigger exposure.
What the research shows

BCG finds pilots often fail to expand company-wide and recommends concentrating people and capital on a few high-value areas, funded over several years. BCG, Jul 2026

Have rules to stop failing pilots31%
Have criteria to retire low-value AI in production17%

In Gartner's survey of 2,002 CIOs, only 13% report significant AI value. Those with clear stop rules for pilots were 2.3 times more likely to be among the top performers. Gartner, Oct 2026

What teams can do

Whichever way you lean, agree in advance what result earns more funding and what result ends an initiative.

4  Board view and reallocation

Showing the board the outcomes its investments enabled is a win in itself, and it is what makes moving money possible.

What I hear

One data leader took the board through the operational outcomes its technology investments had enabled, at a level of detail it had not seen before. That visibility alone was received as a success.

Where savings are freed up, or where a region competes with group priorities, the hard part is making a case that holds up against investments built a different way.

Investment viewSpend, milestones and status, program by program.
Outcome viewThe operational results each investment enabled, in enough detail to see where the value came from.
What the research shows
81%of AI initiatives showed positive returns at high performers, the companies that track returns continuously, manage AI as a portfolio and move money away from what underperforms. Low performers did not know the return on 29% of theirs. Gartner, Sep 2026

Gartner advises tracking every dollar by outcome category, such as productivity, revenue growth or risk reduction, so investments can be defended and money moved quickly when projects underperform. Gartner, Sep 2026

44% of companies plan to fund generative and agentic AI from the savings of earlier automation programs, savings that have consistently come in below target. Bain's advice: check what the last program actually returned before approving the next, and size the new investment to that. Bain, Jun 2026

What teams can do

Give the board one consistent view of the portfolio by outcome, so funds can move between programs, units or regions on evidence, and savings can be put to work where they return most.

5  After approval

Funded projects are executed and managed well. Whether they delivered the projected value often goes unmeasured.

What I hear
  • Approval, not proofThe funding process requires approval. It does not require evidence that the value arrived.
  • Lessons unusedActual results rarely improve the next case, so tracking feels like overhead.
  • Reporting feels riskyShowing actuals against projections built on limited detail can feel like exposure.
  • Still manualEach measurement step competes with time already committed to delivery.
What the research shows
1 in 5organizations calculate the verified value of AI outcomes against their full cost. Capgemini, Nov 2026

76% of organizations have made, or plan to make, their transformation commitments public, which puts delivery in front of boards and investors. Only 24% account for value leakage when they set the targets. Deloitte, Sep 2026

BCG recommends rigorous post-mortems on major capital projects, so each allocation decision is better informed than the last. BCG, Aug 2026

What teams can do

RecognitionUse measured outcomes to tie the team's work to company goals and recognize it.

ReinvestmentWhen a project beats its target, let part of the extra value fund the team's next priority.

LearningRecord the assumptions behind each case, so a miss shows which one changed and the next case improves.

For the next edition

Who owns value after go-live?

The next edition looks at measurement after go-live: who owns it, what gets in the way, and which incentives lead teams to report actuals against the plan. If your organization has made this work, or learned from where it did not, your experience would help shape it.

Compare notes with Swami

Sources

  1. Deloitte, 2026 Enterprise Cost Transformation Survey, September 14, 2026. 500 senior leaders; more than half at public companies with revenue above $5 billion.
  2. BCG, CEOs see cost and revenue benefits from AI but struggle to scale, July 22, 2026. Survey of 152 CEOs at companies with $500M+ revenue.
  3. Gartner, Survey finds only 22% of organizations have scaled AI across multiple business units, September 1, 2026. 1,303 respondents.
  4. Gartner, 45% of CFOs say their AI investments lean toward productivity, July 20, 2026. 204 finance leaders.
  5. Gartner, Only 13% of CIOs say their enterprise has realized significant or transformational value from AI, October 19, 2026. 2,002 CIOs and technology executives.
  6. Bain & Company, Your AI Budget Is Growing. Your Returns Aren't. Here's Why., June 1, 2026. Automation and AI Pathfinder Survey, 951 companies.
  7. BCG, Capital Allocation Takes More Than Good Instincts, August 25, 2026. Nearly 600 listed multi-segment companies, 2015 to 2024.
  8. Capgemini Research Institute, Nearly six in ten faced significantly higher AI costs than expected, November 12, 2026. 500 senior executives.

Field observations come from conversations with IT, PMO, data and transformation leaders and are shared without names. Figures are as published by each source; follow the links for methods and definitions.

See how Arivue builds a case