CFO Index - Q3 2026

Table of Contents

Key takeaways

83% of firms are overspending on AI, with average spending coming in ~27% above expectations.
AI confidence remains remarkably high: average ROI confidence is 8.2/10, despite widespread budget overruns.
AI is reshaping jobs, not simply replacing them: 32% increased finance headcount because of AI, versus 20% that reduced it.
1 in 3 firms are already reducing hiring for roles that AI can cover, pointing to a bigger impact on future roles than existing jobs.
Finance teams spend ~32 hours every month on executive reporting, and the most AI-mature teams actually spend more time, not less.
The UK is bucking the trend: 70% of UK finance leaders say the change in government leadership increased their confidence, while expected revenue growth reached 13.2%.

Ask AI for a summary of this Pigment page

Already, almost unbelievably, we are at the end of Q3 2026. And while we haven’t seen anything quite as seismic as we did in Q2, there has been plenty to keep CFOs up at night.

The Fed hiked rates for the first time since 2023, oil is above $100 a barrel, and the midterms are just around the corner.

The UK has its seventh prime minister since the Brexit vote, French public debt is at its highest level since 1946, and Merz’s government in Germany is in crisis.

Meanwhile, AI capex has climbed even higher, driven in large part by a switch to consumption-based pricing by major vendors like Anthropic.

So how is the global finance community reacting?

Drivers of uncertainty and response to turbulence

There's no single dominant driver of uncertainty in Q3. All 11 score between 2.76 and 2.99 on a 4-point impact scale.

Every one is rated at about "moderate impact" - CFOs aren't worried about one big thing, they’re just generally worried.

Two lists of drivers of planning uncertainty, scored 1 to 4, all close to "moderate impact" (3). Rose since Q2: data quality 2.87 (+0.06), regulatory changes 2.85 (+0.04), internal alignment 2.76 (+0.04), supply chain 2.92 (+0.02), technological change 2.92 (+0.02), talent 2.81 (+0.02). Eased: geopolitical conflict 2.95 (−0.06), competitive threats 2.79 (−0.04), macroeconomic conditions 2.99 (−0.02), energy costs 2.94 (−0.02). Government leadership or policy changes, new in Q3, scored 2.84.

External shocks (macro, geopolitics, competition) have eased slightly, while internal pressures (data quality, strategic alignment, talent, regulation) have all increased slightly. But geopolitical conflict remains the number one driver of uncertainty globally - for obvious reasons.

Meanwhile, we’re seeing a drop across the board in ‘emergency’ turbulence response activity: it’s been a less busy quarter for finance teams.

Bar chart of actions organizations took in response to turbulence, Q3 vs Q2 2026. Re-forecasting plans 51.2% (down 5.1 points), additional scenario plans 48.2% (down 2.1), higher cash reserves 47.0% (down 0.7), revised supplier strategy 22.0% (down 4.6). New this quarter: accelerated strategic decisions 42.1%, accelerated hiring 29.6%, delayed hiring 26.7%, delayed strategic decisions 20.6%. None of the above: 2.9%.

Interested in improving your scenario planning capabilities? We’ve got you covered - download our playbook here.

Revenue performance, expected growth, and confidence

Revenue performance in the past year, in the US and in the UK, has remained resilient even in the face of uncertainty. France and Germany meanwhile are significantly gloomier, both posting decreases from their Q2 figures.

In the growth forecasts, the UK is the only market improving, with respondents predicting 13.2% higher revenues in the year to come, up 0.41pp from Q2.

The same story plays out in the confidence figures - we've seen dramatic drops in the confidence levels (share saying they're moderately or very confident in the direction of their organization) in the US, France, and Germany.

But in the UK, that number has increased 2.1pp, which means it's now the most optimistic market.

Among VP/CFOs and directors, "very confident" fell about 9 points, while among managers it was almost unchanged (+0.8).

The people closest to the board, the cost of capital and the full P&L lost the most confidence - perhaps unsurprising in a quarter in which rates rose and growth assumptions had to be revisited.

The Burnham bounce, and the midterms loom large

It’s official: we’re seeing a ‘Burnham bounce’ in the UK.

70% said the change in the UK government leadership had directly increased their confidence (26% significantly), and only 9% said it decreased. Taken with the growth forecasts (and the fact that the UK is the only market showing an increase in planned AI investment numbers), the situation there is looking comparatively rosy.

Meanwhile, the US midterms are a big concern for finance teams globally.

Of course it’s most intense in the US (48% say it’s fuelling uncertainty in their planning to a great/very great extent), but in Germany that number is 41%, 34% in the UK, and 24% in France.

And the French 2027 election is already weighing on plans. Of French respondents, 40% say it adds uncertainty to a great extent or more, rising to 54% among VP/CFOs.

Is AI out of control?

Most now have AI governance in place

The majority of firms say they control their AI tools centrally.

Single stacked bar of how organizations govern AI tools, from least to most control: no formal controls 3.9%, informal guidelines 15.3%, approved tools 25.1%, centrally managed 28.8%, comprehensive controls 26.2%. In total, 19% have informal or no controls and 55% manage AI centrally or comprehensively.

Over half (55%) maintain either centrally managed (29%) or comprehensive (26%) controls. A quarter (25%) have defined approved tools, and 19% have only informal guidelines or no controls.

But a familiar trend from past quarters is playing out in the data again: leaders and managers see governance very differently.

Of VP/CFOs, 66% describe centrally managed or comprehensive controls, against 45% of managers. A quarter of managers (25%) say governance is informal or non-existent.

Looking for a way to understand your spend? Read about Pigment’s AI Investment Planner here.

It seems that the policies enacted by senior management may not be penetrating through organizations properly yet. That’s a serious problem, because…

Everyone is overspending, badly

83% say consumption-based costs have pushed spending above expectations. Only 13% have stayed within budget, and 4% can't measure their spend at all. The median overrun is 11–25%.

However, 38% are more than 25% over and 17% are more than 50% over, which brings the average to around 27%.

Column chart of how far AI spending exceeded expectations due to usage-based costs. Not exceeded 13.0%; 1–10% over 18.1%; 11–25% 26.7%; 25–50% 21.8%; 51–75% 12.3%; 75–100% 3.4%; more than 100% 0.9%; can't measure 3.8%. In total, 83% are over budget, 38% by more than 25%, and the average overspend is about 27%.

We’re seeing the impact of this in the AI investment numbers: the expected increase in budgets for the next year dropped from 19.1% to 15.6%, the lowest yet. Crucially, that’s lower than the average overspend this year - a clear sign AI spending growth has peaked.

But they still think it’s worth it

Confidence in the ROI of AI is high overall. The average score is 8.2 out of 10, with 50% scoring 9–10 and only 12% scoring 6 or below. This is a little bit odd because ‘unclear ROI’ is a growing barrier across quarters.

Column chart of confidence that AI is delivering meaningful ROI, on a scale of 1 to 10. Shares by score, 1 to 10: 0.6%, 0.3%, 0.9%, 1.2%, 3.0%, 6.2%, 13.6%, 24.5%, 25.0%, 24.7%. Scores 1–6 total 12.2%, 7–8 total 38.0% and 9–10 total 49.7%. The average is 8.2.

Benefits and barriers

The benefits being derived from AI in finance teams are, once again, shifting. Greater productivity and efficiency is still the largest perceived benefit, but it’s down 5 points this quarter.

Reduced manual work, which one might expect to be one of the largest benefits, has declined every quarter so far.

Paired bar chart of how AI delivers value to finance teams, Q3 vs Q2 2026. Productivity and efficiency 54.7% (down 5.0 points), faster planning 50.4% (down 3.6), accuracy and confidence in planning 49.7% (up 3.0, the only benefit to rise), quality of decision making 49.6% (down 1.2), more time for strategic projects 42.3% (down 0.1), reduced manual work 41.3% (down 1.4), improved work experience 29.7% (down 0.7), unsure or no impact yet 0.9% (down 0.7).

But accuracy and confidence in planning outcomes have risen QoQ, now drawing almost level with productivity and efficiency - as workflows become more optimized and engrained, teams are able to plan better rather than just faster.

Looking at the barriers, we can see that the skills gap is closing, but data quality remains a sticky issue.

Paired bar chart of barriers to greater AI use in planning and forecasting, Q3 vs Q2 2026. Data quality or availability 40.4% (up 0.1 points, top barrier), high costs 36.1% (down 2.9), unclear ROI 27.6% (up 3.2), regulatory or security concerns 26.1% (down 2.7), lack of internal skills 25.0% (down 5.4, the biggest fall), slow procurement or implementation 22.9% (up 0.8), cultural resistance 19.9% (down 0.7), lack of leadership buy-in 19.7% (down 0.2), poor quality outputs 11.1% (down 1.3), no barriers 14.3% (up 2.9).

Learn how to improve your data quality posture in our playbook, Operationalizing AI in finance.

Cost is a smaller barrier than you'd expect given how many firms are over budget. Only 36% cite high costs as a barrier, down three points, even though 83% of AI users are over budget. But this does make sense intuitively - dramatic overspending clearly hasn’t been curtailed to this point, so it’s unlikely to show up as a barrier for finance teams looking to derive value from it.

As governance policies tighten, it’s likely we’ll see that number increase.

Impact on the office of finance

Where AI is exceeding expectations

Respondents are incredibly bullish on AI. For every task we asked about, more respondents said AI was ‘more useful’ than expected than said it was ‘less useful’.

Unsurprisingly, AI is least successful for the tasks that demand human judgement: scenario planning and model building, for example.

Diverging bar chart of the share saying AI has been more or less useful than expected, by finance task, with net score. Budgeting 49.9% more vs 7.8% less (net +42), reports and summaries 48.1% vs 10.2% (+38), forecasting 44.2% vs 10.5% (+34), data classification 43.9% vs 11.9% (+32), internal reporting 42.1% vs 11.5% (+31), variance analysis 41.0% vs 12.6% (+28), model building 38.7% vs 11.4% (+27), scenario planning 39.9% vs 14.3% (+26).

Reshaping rather than shrinking

More firms say AI has increased their finance headcount (32%) than reduced it (20%). The most common change is to the work itself: 36% have changed the scope of existing roles, and 28% have created new finance roles. Only 12% say AI hasn't yet led to significant changes.

Bar chart of how AI has affected finance roles, staffing and structure. Changed scope of existing roles 35.9%, reduced hiring 32.5%, increased headcount 31.5%, redesigned or new roles 27.7%, hired AI or data specialists 27.7%, shifted work to AI 23.5%, more training or reskilling 22.8%, reduced headcount 19.7%, more external consultants 17.0%, merged teams 15.3%, shifted work between functions 14.2%, no significant change 11.6%.

But a third of firms (33%) have reduced hiring for roles AI can now cover, and that figure is between 30% and 35% in almost every country, company size and seniority level we looked at. AI isn't cutting many existing finance jobs, but it's reducing the number of new ones - the entry-level roles, where future finance leaders get their start.

There's also a gap in reskilling. More than a third of firms have changed what their finance staff is responsible for, but fewer than a quarter have increased training.

The executive reporting burden

Finance teams spend a lot of time producing reports for leadership. The largest group (39%) spends 21–40 hours a month preparing dashboards and reports for executives, and one in five (21%) spends more than 40 hours.

Grouped column chart of monthly hours the finance team spends on executive reporting, all respondents vs enterprises (20,000+ employees). Less than 5 hours: 1.6% vs 3.3%; 5–10: 10.1% vs 10.5%; 11–20: 28.0% vs 21.0%; 21–40: 39.0% vs 25.7%; 41–80: 15.0% vs 21.4%; more than 80: 5.6% vs 17.6%. More than 40 hours applies to 21% of all teams and 39% of enterprises. The typical team spends about 32 hours a month.

That works out to roughly 32 hours a month for the typical team, or around ten working weeks a year. At the enterprise level, it’s an even bleaker picture: 39% spend more than 40 hours a month, and 18% spend more than 80.

Most surprising, AI doesn’t seem to be reducing it. Automatically generating reports and summaries is one of the tasks where AI has most exceeded expectations (48% say it's more useful than expected).

But 30% of the most AI-mature firms spend more than 40 hours a month on executive reporting, compared with 17% of early-stage firms. Perhaps when reports become quicker to produce, leadership asks for more of them.

If only there was a way to instantly generate reports for leadership… Read up on Pigment Frames here.

Join us to discuss the findings

Too much to take in at once?

Pigment’s incoming CFO, Susan Phan, will be hosting a webinar with John Van Decker from Dresner Advisory Services, to discuss the findings, what they mean for you, and how they should inform your plans.