AI Productivity Gains Are the Next Earnings Story: Why FP&A Could Become the Most Important Team in the Room

Investors are beginning to look beyond AI announcements and ask which companies are turning artificial intelligence into measurable earnings growth. That’s an opportunity for FP&A to do something far more valuable than track technology spending: help management discover where AI is improving the business and where to invest next.

Apparently, telling Wall Street that your company is using artificial intelligence is no longer enough.

This is disappointing news for executives who spent the last three years adding the phrase AI-powered to presentations.

Some of those presentations were very attractive.

Beautiful graphics.

Ambitious transformation roadmaps.

And enough references to productivity to suggest that the entire workforce would soon finish its assignments before breakfast.

Unfortunately, investors have started asking an awkward question.

Where’s the money?

Not the money being spent on AI.

We’ve heard plenty about that.

They want to know where the additional profit is showing up.

And I think that’s excellent news for FP&A.

Because measuring financial performance is something we already know how to do.

We just haven’t always been invited into the technology conversation early enough.

On October 13, 2026, Business Insider reported that investors are increasingly looking for evidence of AI-driven productivity improvements as third-quarter earnings season begins.

The article cited research indicating that approximately half of S&P 500 companies mentioned AI during second-quarter earnings calls.

Yet only 11% discussed specific use cases.

And just 2% quantified the earnings impact.

Think about that.

AI is apparently everywhere.

Except, in most earnings discussions, the financial bridge explaining what it actually accomplished.

I don’t see that as proof that AI isn’t working.

I see it as evidence that companies have an opportunity to become much better at measuring the results.

And the organizations that figure it out may have an advantage.

Why Investors Are Asking for AI Productivity Numbers

For several years, much of the investment discussion surrounding AI focused on the companies building the infrastructure.

Semiconductors.

Cloud computing.

Data centers.

Networking equipment.

Energy systems.

Those businesses are supplying the technology required to support AI development.

But another opportunity is emerging.

Companies that use AI to improve their existing operations.

Retailers that manage inventory more efficiently.

Healthcare businesses that reduce administrative work.

Financial institutions that improve risk analysis.

Manufacturers that identify production problems earlier.

Professional services firms that complete routine work faster.

The investment case for these companies is different.

They’re not necessarily selling AI technology.

They’re using it to improve the economics of businesses they already operate.

Morgan Stanley Wealth Management chief economist Ellen Zentner has argued that the next phase of AI investment could favor companies using the technology to improve decisions, reduce costs and strengthen margins.

I find that direction compelling.

Because the economic value of AI doesn’t have to be concentrated in the companies developing the technology.

It can also emerge in the businesses applying it effectively.

But to recognize that value, companies need to measure it.

And that requires more than counting how many employees have access to an AI tool.

The Difference Between AI Adoption and AI Productivity

Let’s establish something important.

Using AI isn’t the same as becoming more productive.

An employee might use AI to draft a report.

That doesn’t automatically mean the company saved money.

The employee may spend additional time checking the output.

The report may require revisions.

Or the time saved may simply be absorbed by other activities.

That doesn’t make the technology useless.

It means adoption and financial value are different measures.

I would separate AI performance into several levels.

The first is adoption.

How many people are using the technology?

The second is operational productivity.

How much time, effort or processing capacity does it save?

The third is financial impact.

Does the improvement reduce expenses, increase revenue, avoid future costs or improve working capital?

And the fourth is strategic value.

Does the technology enable the company to do something it couldn’t do previously?

Those are different outcomes.

A successful AI implementation might deliver value in one or several categories.

But I’d want management to be clear about which benefits it’s expecting.

Because a company can have impressive adoption statistics and very little measurable financial return.

And it can also have a relatively small AI implementation that produces substantial value.

The number of users isn’t necessarily the most important number.

The $500,000 AI Productivity Question

Let’s imagine a finance organization with 100 employees.

Management introduces an AI tool that saves each employee an average of five hours per month.

That’s 500 hours saved monthly.

Or 6,000 hours annually.

Assume the fully loaded employee cost is $80 per hour.

The theoretical annual labor capacity released is:

6,000 hours × $80 = $480,000.

That sounds wonderful.

Nearly half a million dollars in productivity.

Someone will undoubtedly put that number in a presentation.

But I wouldn’t call it $480,000 in realized cost savings.

Not yet.

The employees are still employed.

Their salaries haven’t changed.

And the company may not have reduced any external spending.

What it has created is additional capacity.

That’s valuable.

But the financial benefit depends on what happens next.

Perhaps the company avoids hiring three additional analysts.

Perhaps the employees use the time to support more customers.

Perhaps the finance team produces better forecasts.

Or perhaps everyone spends more time in meetings.

I sincerely hope it’s not the last one.

The point is that time saved is an operational benefit; money saved is a financial outcome.

And finance needs to explain the relationship between them.

How FP&A Should Calculate AI Return on Investment

I would build an AI investment case using three categories of benefits. For a fuller framework, see how to build an AI business case for finance.

First, realized financial savings.

These include expenses the company has actually reduced or avoided.

Second, measurable operating improvements.

These include productivity gains that create capacity but haven’t yet translated into lower costs or additional revenue.

Third, strategic benefits.

These might include better decisions, improved customer service or the ability to develop new products.

All three matter.

But they shouldn’t be combined into one enormous savings number without explanation.

Let’s consider a hypothetical AI implementation.

A company spends $300,000 annually on software, implementation support and ongoing maintenance.

It expects the technology to improve customer service operations.

The estimated benefits are:

Benefit	Estimated annual value
Reduced outsourced service costs	$180,000
Avoided incremental hiring	$150,000
Lower rework and error costs	$70,000
Total measurable financial benefits	$400,000
Annual AI-related costs	($300,000)
Estimated net annual benefit	$100,000

Under those assumptions, the annual benefit-to-cost ratio is approximately 1.33.

That’s a positive result.

But I’d want to understand when the benefits begin.

If implementation takes six months, the first-year economics may be different.

I’d also examine one-time integration costs, training expenses and the possibility that actual usage is lower than expected.

And I’d make sure the avoided hiring estimate is supported by an actual staffing plan.

If the company wasn’t going to hire those employees anyway, the benefit is questionable.

Finance should be able to explain that distinction.

Preferably before the number appears in an investor presentation.

Why AI Margin Improvement Could Matter More Than AI Revenue

A lot of the AI conversation focuses on new revenue opportunities.

Those are important.

But I think operating margin improvement may become one of the most compelling near-term measures of AI value.

Consider a company with $500 million in annual revenue.

Its operating margin is 12%.

That produces $60 million in operating profit.

Now suppose AI-enabled process improvements reduce annual operating expenses by $5 million, with no change in revenue.

Operating profit increases to $65 million.

The operating margin improves from 12% to 13%.

That’s a one-percentage-point improvement.

And an 8.3% increase in operating profit.

The company hasn’t entered a new market.

It hasn’t launched a new product.

It hasn’t increased prices.

It has improved its operating efficiency.

That’s a meaningful result.

Of course, the savings need to be real.

If the company spends $7 million to generate $5 million in annual savings, the financial picture changes.

That’s why I’d evaluate the full investment economics.

But the underlying opportunity is substantial.

For businesses with large operating cost bases, relatively small efficiency improvements can create meaningful earnings growth.

And AI may help companies identify those opportunities.

The Most Valuable AI Use Cases May Be Surprisingly Ordinary

I suspect some of the most profitable AI applications won’t make particularly exciting headlines.

They may involve invoice processing.

Customer service routing.

Inventory forecasting.

Document review.

Maintenance scheduling.

Collections prioritization.

Or reconciling information across systems.

Not exactly the material that inspires dramatic technology conferences.

But these processes involve real costs.

And they often contain measurable inefficiencies.

Let’s take invoice processing.

Suppose a company processes 200,000 supplier invoices annually.

Its current fully loaded processing cost is $8 per invoice.

That’s $1.6 million annually.

Now suppose automation reduces the average cost to $5.

The potential gross savings are:

200,000 × $3 = $600,000 annually.

If the technology costs $200,000 per year to operate, the estimated net annual benefit is $400,000.

That’s worth examining.

But I’d also want to know whether the lower processing cost reflects actual expense reductions or simply less staff time per invoice.

If the company redeploys employees rather than reducing costs, the financial benefit needs to be measured differently.

The implementation may still be valuable.

It may allow the company to process higher volumes without increasing headcount.

That’s a legitimate operating advantage.

But it should be described accurately.

Why FP&A Should Own the AI Benefits Measurement Framework

I don’t think FP&A needs to own every AI project.

Technology teams should manage technical architecture.

Operations should understand workflow requirements.

Risk and compliance teams should establish appropriate controls.

But finance should help establish how financial benefits are measured.

Otherwise, companies can end up with several departments reporting overlapping savings.

The technology team reports 10,000 hours saved.

Operations reports improved productivity.

Human resources reports avoided hiring.

And finance reports lower projected labor costs.

All four benefits may originate from the same underlying change.

If we add them together without reconciling the assumptions, we can count the same economic value several times.

That’s not a productivity improvement.

That’s creative arithmetic.

I’d establish a common measurement framework.

Each project would identify the operational baseline.

The expected improvement.

The financial conversion method.

The implementation and ongoing costs.

The timing of benefits.

And the person responsible for verifying results.

That doesn’t need to become an enormous administrative exercise.

It needs to be consistent enough that management can compare investments.

A Practical AI Benefits Scorecard for CFOs

Here’s the type of scorecard I’d want to see.

Metric	Why it matters
Process volume	Establishes the scale of the opportunity
Cost per transaction	Measures unit economics
Hours per task	Tracks operational productivity
Error or rework rate	Captures quality-related costs
Actual expense reduction	Identifies realized savings
Avoided future spending	Measures supported cost avoidance
Incremental revenue	Captures growth benefits
AI operating cost	Reflects ongoing investment
Net financial benefit	Shows economic contribution
Benefit realization date	Distinguishes planned from actual results

I would also include quality measures.

An AI system that processes invoices faster but introduces more errors isn’t necessarily an improvement.

Likewise, a customer service tool that reduces handling time while frustrating customers may create downstream costs.

The objective is better business performance.

Not simply faster activity.

And I’d want the scorecard to show both.

Why the Next Earnings Season Could Change AI Reporting

Business Insider’s October 13 report suggests investors are becoming more interested in which companies can demonstrate measurable AI benefits.

I think that makes sense.

During the early stages of a technology cycle, companies often focus on investment and adoption.

Management explains what it’s building.

How much it plans to spend.

And what the technology might eventually accomplish.

As implementation progresses, the conversation changes.

Investors begin asking about results.

Did costs decline?

Did productivity improve?

Did margins expand?

Did revenue grow?

Was the return worth the investment?

That’s a natural progression.

And I think it could be healthy for the market.

Companies that can demonstrate financial improvements may distinguish themselves from those offering only ambitious projections.

But I wouldn’t assume every company needs to publish a precise AI earnings contribution immediately.

Some benefits are difficult to isolate.

Others take time to emerge.

The important thing is developing credible measurement.

Because the companies that understand their own results will be better positioned to explain them.

AI Could Improve Financial Decisions, Not Just Reduce Costs

There’s another part of this story that I think deserves more attention.

AI may create value by helping companies make better decisions.

Not simply by automating work.

Consider inventory management.

A company might use AI to improve demand forecasting.

That could reduce excess inventory.

Lower storage costs.

Improve product availability.

And reduce markdowns.

The benefit isn’t necessarily fewer employees.

It’s better allocation of working capital.

Suppose a retailer reduces average inventory by $10 million without harming service levels.

That releases $10 million in working capital.

The company could use the cash to reduce borrowing, fund expansion or strengthen liquidity.

If the inventory reduction is sustainable, it may also reduce carrying costs.

That’s a financial benefit.

And potentially a substantial one.

But I’d distinguish between the cash released and the recurring income statement savings.

They’re not the same thing.

The company shouldn’t count a $10 million working capital release as $10 million of annual profit.

I know.

Finance ruins all the fun.

But accurate measurement is what makes the result credible.

The Difference Between AI Cost Savings and AI Growth Capacity

One of the most interesting opportunities is using AI to support growth without increasing expenses at the same rate.

Imagine a professional services business.

It currently generates $50 million in annual revenue with 200 employees.

Management expects demand to increase by 20%.

Under the existing operating model, that might require additional hiring.

But suppose AI-enabled workflows allow the company to handle more work with its current team.

The business may be able to increase revenue without proportionately increasing headcount.

That’s operating leverage.

And it could be extremely valuable.

Let’s create a simplified example.

Assume the company generates $50 million in revenue and $8 million in operating profit.

Now suppose revenue increases to $60 million.

Variable delivery costs rise by $4 million.

Other operating expenses increase by $1 million.

Operating profit becomes $13 million.

That’s a 62.5% increase in operating profit on 20% revenue growth.

AI may contribute to that result by increasing productive capacity.

But I’d want to understand how much of the improvement is actually attributable to the technology.

Perhaps pricing improved.

Perhaps the customer mix changed.

Perhaps demand was stronger than expected.

Finance needs to separate those drivers.

Otherwise, every favorable variance risks becoming an AI success story.

And that would make earnings calls considerably more entertaining.

How to Avoid Double-Counting AI Productivity Gains

This is where measurement can become tricky.

Suppose an AI tool saves 5,000 employee hours annually.

Management estimates those hours are worth $400,000.

Then the company uses the additional capacity to generate $600,000 in incremental revenue.

Should finance report $1 million in benefits?

Not automatically.

The same hours may have enabled the additional revenue.

If the company counts both the theoretical labor savings and the resulting contribution from new business, it may be double-counting the benefit.

I’d want to understand the economics.

If the hours saved allow employees to serve additional customers, the financial benefit may be the incremental contribution margin from that revenue.

If the company instead reduces staffing costs, the benefit may be the actual expense reduction.

If it does both through separate capacity changes, both could be valid.

But the model needs to demonstrate that.

This is why AI benefit measurement requires financial discipline.

Not because the technology isn’t valuable.

Because we want to understand precisely how the value is created.

A Better Way to Build an AI Investment Portfolio

I think companies should evaluate AI initiatives as a portfolio of investments.

Some projects will produce relatively predictable efficiency improvements.

Others will create new revenue opportunities.

Some may improve risk management.

And some will be experimental.

That’s normal.

The challenge is deciding where to allocate capital.

I’d begin by comparing projects using consistent financial assumptions.

Expected cost.

Potential benefit.

Time to implementation.

Probability of success.

Operational complexity.

And strategic importance.

A project that saves $100,000 annually with minimal implementation risk may be attractive.

A project that could generate $5 million in new revenue but requires significant development may also be worthwhile.

They serve different purposes.

The company doesn’t need to choose only one type.

But management should understand the trade-offs.

And FP&A can help establish the financial framework.

That’s an opportunity to become more involved in technology strategy.

Not by becoming software engineers.

By making sure investment decisions reflect credible economics.

Why AI Productivity Reporting Should Be Connected to the Forecast

One thing I would change in many organizations is the separation between AI project reporting and financial forecasting.

Technology teams often track implementation milestones.

Operations tracks productivity.

Finance tracks expenses.

But those measures don’t always connect.

A project may report that it has delivered its expected productivity improvements.

Yet the financial forecast remains unchanged.

That should prompt a conversation.

Where is the benefit appearing?

Has the company reduced spending?

Avoided planned hiring?

Increased output?

Improved margins?

Or created additional capacity that hasn’t yet been used?

Those are all possible outcomes.

But management needs to know which one occurred.

I’d want AI project benefits linked to the financial forecast.

If an initiative is expected to reduce costs beginning in January, the forecast should reflect the timing and amount.

If implementation is delayed, the expected savings should move.

If the project exceeds expectations, management should evaluate whether additional investment is justified.

That creates accountability.

And it makes the forecast more useful.

What I’d Want Companies to Disclose About AI Productivity

I don’t expect companies to reveal every internal process or competitive advantage.

But I do think investors would benefit from more specific information.

For example, a company might explain that AI-enabled automation reduced the cost of processing customer requests.

Or that improved demand forecasting lowered inventory requirements.

Or that software development productivity increased enough to support additional products without proportional staffing growth.

The company could provide a measurable operating indicator.

And, where practical, explain the financial effect.

That would be more useful than another general statement about AI transformation.

I also think management should distinguish realized benefits from expected future benefits.

A company may have an excellent AI strategy.

But the value of an implemented project is different from the value of a planned initiative.

Investors understand that distinction.

Finance should too.

The Next AI Winners May Be Companies Nobody Thinks of as AI Companies

This is the part of the story I find most exciting.

The next major beneficiaries of AI productivity may not be the companies building the models.

They may be established businesses with large operating cost bases and substantial opportunities for efficiency improvements.

A retailer with complex inventory requirements.

A manufacturer with expensive maintenance processes.

A healthcare company with administrative workflows.

A logistics business managing thousands of daily decisions.

Or a financial institution processing enormous volumes of transactions.

These companies already have customers.

They already have revenue.

And they already have operating processes that could potentially be improved.

If AI helps them reduce costs, improve decisions or expand capacity, the financial benefits may be meaningful.

That’s an attractive investment story.

But it’s also a management story.

The companies that identify the right use cases, implement them effectively and measure the results may gain an advantage.

And I think FP&A can play an important role in making that happen.

FP&A Has an Opportunity to Become the AI Value Office

I wouldn’t create another department simply because we need a new acronym.

Corporate organizations have enough of those.

But I do think finance should establish a clear process for evaluating AI investments and tracking their financial results.

The technology team can explain what the system does.

Operations can explain how the workflow changes.

FP&A can explain what the change means financially.

That combination is powerful.

It allows management to move beyond adoption statistics and understand actual economic value.

It also helps companies identify which projects deserve additional investment.

Because once an AI initiative demonstrates measurable benefits, the next question becomes obvious.

Can we apply the same approach elsewhere?

That’s where the opportunity grows.

A successful project becomes a repeatable operating improvement.

And the financial model helps determine whether scaling it makes sense.

I Think the AI Earnings Conversation Is Finally Getting Interesting

For a long time, the AI investment discussion has focused on how much companies are spending.

That was an important first chapter.

But I think the next chapter could be more interesting.

Which companies are improving margins?

Which are releasing working capital?

Which are increasing revenue without proportional cost growth?

And which are using AI to make better decisions?

Those are questions finance can help answer.

I don’t see investors demanding evidence of productivity as a negative development.

I see it as an opportunity.

Companies that can demonstrate measurable improvements may attract more confidence.

Management teams may allocate capital more effectively.

And FP&A may become more involved in decisions that shape the business.

That sounds like progress to me.

Of course, it means we’ll need to develop better financial models.

Establish baselines.

Track results.

And occasionally explain that saving 10,000 employee hours doesn’t automatically mean the company saved $800,000.

But that’s the work.

The exciting part isn’t that AI can generate another report.

It’s that AI may help companies become more profitable.

And finance has an opportunity to prove it.

Personally, I’m looking forward to the day an earnings call includes a detailed explanation of AI productivity gains.

Actual cost savings.

Improved operating margins.

Better working capital.

Clear financial returns.

No vague references to transformation.

No mysterious productivity multipliers.

Just a credible explanation of what changed and how much value it created.

Although I suppose that would leave executives with considerably fewer opportunities to say AI-powered.

I’m willing to make that sacrifice.