Value Engineering vs AI Cost Cutting

In the Episode

When AI spending lands in the operating budget, the first thing many companies cut is people. This episode explains why that happens and what to do instead.

• Why AI costs end up as headcount cuts: AI token spend is an operating expense, usually charged to IT, and labor is the easiest line to cut once capital and SaaS commitments are locked in.
• Cost myopia: cost is right in front of you, and value is further away. Leaders focus on cost because it’s easier to control and because cause and effect between investment and value is hard to prove.
• Value engineering, explained: a practice that began at GE in the 1940s. It breaks a capability down into what its parts cost and the value each part produces. Value is benefits minus cost per unit of time.
• What chargebacks hide: centralized infrastructure costs left IT spend understated by 40–60%, until a corporate split forced every dollar to land in someone’s budget.
• Avoiding the wrong investment: a value engineering analysis prevented a $20M platform spend and found that the real problem was back orders.
• Measuring what seems unmeasurable: if you find the right indicator, you can measure things that look intangible.
• Where to start: change is driven by pain more often than by vision. Fix what’s in your control first, then deliver value in 90-day vertical slices that fund the next step.

Frequently Asked Questions

Why does AI cost cutting lead to layoffs instead of smarter IT budget decisions?

AI spend, especially tokens, is an operating expense, and it usually lands in IT. Capital spend and SaaS contracts are locked in, so labor is the easiest line left to cut. The real problem is cost myopia: leaders manage the cost in front of them because value is harder to see and harder to prove.

What is value engineering, and how does it measure business value?

It’s a discipline that began at GE in the 1940s. You break a capability, process or value stream into what each part costs and the value it contributes. The basic test is simple: value is benefits minus cost per unit of time.

How do IT chargebacks and cost-only budgeting hide the true cost of technology?

Chargebacks move costs around without tying them to value, so nobody owns the real number. In one company, chargebacks hid enough infrastructure spend that IT costs were understated by 40–60%. In another, a value engineering analysis stopped a $20M platform investment and pointed the money at the real problem, which was order fulfillment.

Can internal teams run a value engineering analysis and measure intangible benefits?

Yes. These skills already exist in finance and engineering teams. Almost anything can be measured once you find the right indicator.

Will value engineering prevent AI-driven layoffs and improve AI ROI?

Not on its own, but better math is better than no math. Financial models keep you from making the wrong cuts and the wrong investments, and they keep productivity gains from disappearing without ever turning into value.

How do leaders move from cost cutting to value-based decision making?

Not by starting with culture, because culture is an effect, not a cause. Start where the pain is and fix what’s in your own control first. Then deliver value in 90-day slices, and let each one fund the next.

Podcast Transcript

Len Greski
Executives at the Museum of Science and Industry in Chicago wanted to figure out what the most popular exhibit was. And they sent out a survey. People were likely to mention snobbish exhibits that were scientifically complicated. If you actually looked at those locations in the museum, there were not many people that typically frequented those exhibits.

They asked some sociologists at the University of Chicago how they’d solve this problem. And the answer that came back was, “Let’s go talk to the maintenance engineers and figure out where they replace tiles, floor tiles throughout the building.” And the winner was the baby chick exhibit. And so one could make an argument you can’t measure the most popular exhibit because people will lie to you on a survey.

And the answer was actually you can measure, but you just have to know what’s a better way to measure it.

Stacy Gordon
Hey everybody. Stacy Gordon here, host of the podcast Future State Now. I’m happy to be spending time with Len Greski, chief scientist at LiminalArc. Len, you wrote an amazing article recently called Why Value Engineering Beats AI-Driven Cost Cutting.

And I loved how you opened the article because you talked a lot about reports that are out there today saying AI is eliminating jobs, but I believe you don’t think that’s the right story. What is the right one?

Len Greski
Well, the right story, and it’s been this way for decades related to AI, is about value. And that people have been working in the AI space for 40 years, 50, if you go back to the 1970s applications like MYCIN, which was an early tool for doctors to diagnose illnesses. And at that time, AI or expert systems and neural network applications were targeted at very specific business problems that were trying to be solved where there was a very clear value proposition because at the time everything was custom software.

You couldn’t download a neural network as a Python package, for example. And now AI is so much more accessible that people are focusing on how do you free up costs to experiment with it versus looking at what are we actually going to do and how does what we’re going to do generate value?

Stacy Gordon
And so why does that AI usage bill always seem to end up on headcount?

Len Greski
Well, because if you think about what happens in a company with various cost structures, there’s a big distinction between capital expenditures and operating expenses. And in an earlier article I wrote this year called The New Software Economics, talk about the history of accounting treatment for software development work and how AI enabled development introduces new categories of spend like the dreaded AI tokens.

And those categories of spend are inherently operational because they’re not baked in to the newly created software asset. So companies typically centralize these operational costs in IT because if it looks like IT and it smells like IT, IT going to get the expenses and IT may or may not allocate those expenses out to business units based on some form of rationalization, whether it’s number of employees or number of servers that are used by business unit.

And if the costs aren’t allocated to a business unit, then the IT department gets stuck with them and then it’s got to figure out where else to cut to meet the annual budget commitments. And so if in IT, the labor is neither unionized nor protected by some fixed bid contract, like you’ve outsourced X, you said you’re going to spend 20 million a year on it and it’s the outsourcer’s responsibility, staff the labor, you’re still on the hook for the 20 million. Labor’s the easiest thing to cut.

Stacy Gordon
Right. Is that where you came up with your term? And if you haven’t spent a lot of time with Len, he always will give you what I call Len-isms, things that he brings back from the past to make a harder situation sound easier to understand. But I loved your term cost myopia.

Len Greski
Well, that was something that as I was writing the article, I kind of like to break the text up into chunks with some kind of heading. And that paragraph, that’s what it is. Myopia is a vision problem where the visual images come to a focus in front of the retina and that results in poor vision of distant objects.

And a secondary definition is a lack of foresight or discernment or a narrow view of something. And cost is right in front of you, whereas value is further out. And so myopia is a good way to metaphor for the focus on cost rather than value. And leaders focus on cost rather than value for two reasons.

Number one, it’s easier to control your costs rather than benefits or more specifically revenue. And second is that many companies struggle to establish cause and effect between the investments they make and the value they expect to realize from them.

Stacy Gordon
And is that why you would say labor’s kind of always the relief valve? It’s the easier one to understand?

Len Greski
That’s right because typically a department is going to, a budget is broken out into different categories of expense. And in IT, you’ve got labor and non-labor expenses and then you’ve got capital expenses versus operating expenses. And the things that are easiest to change or it’s easier to change OPEX than CapEx because once you’ve capitalized something, you’re stuck with its amortization schedule.

You can’t undo the capital expenses in some ways. They’re like the gift that keeps on giving. Once you’ve spent it, you’re stuck with it for you. And so then you’re looking at your operating expense budget and then within the operating expense budget, you have labor and non-labor.

Now, over the past 20 years, more non-labor expenses have become operating expenses where it used to be you’d buy a wave of servers for your data center, you’d depreciate them over, you’d put them on a seven year amortization schedule and those costs were fixed for the better part of a decade. While you couldn’t undo them, they were lower than if you were paying operating expenses each year for that compute capacity.

With the rise of software as a service and cloud computing, as I wrote about in the new software economics, those categories of expenses have moved from CapEx to OpEx. And so if you’ve got a commitment of 10 million a year in spend to your SaaS vendor, it’s really hard to reduce that.

Stacy Gordon
Sure.

Len Greski
And so then labor becomes the relief valve.

Stacy Gordon
Yeah. Well, I mean, it’s got to be frustrating. You’ve sat in a number of executive seats throughout your career, but when you get handed at an operating budget that you didn’t get to set and you’ve got to make tough decisions within that, that’s a struggle for anyone.

Len Greski
Well, it is, but it’s also a reality of a business that we all have to face into is the budget is a fixed thing and that’s why it’s so important to be able to understand the value that’s generated by your budget. Because if I don’t understand the value being generated by my budget, I can’t communicate that to stakeholders.

And if I can’t communicate the value, I’m at risk for severe cuts. Because when the senior most executives come looking for budget cuts, the budget monster is out trying to grab as much spare cash as possible, they’ll target a percentage reduction and peanut butter spread it across all the departments. And it’s very difficult to say no when your budget cuts our peanut butter spread and it’s like everybody’s got to give their pound of flesh.

One way that I’ve defended myself against that in other companies is to outsource work and run it on annual contracts so that if nobody wants to pay for an asset, you just shut it off. And in one company, our fiscal year ended on the 31st of October, and I would literally talk to my stakeholders like, “Well, if you don’t want to pay for that anymore, it’ll turn into a pumpkin on Halloween and it won’t be there on all saints day.”

And that would lead to some very difficult conversations with business stakeholders because people are often very excited about building new capability, but nobody wants to pay to operate what you’ve already got in place.

Stacy Gordon
Oh, well, I’ve heard that statement over my career. You’re always more valuable somewhere else than you are internally at times, but I’m sure I’m not the only one. Now, you’ve mentioned a couple times value engineering. Break that down for me.

What does that actually mean to the common person and where did it originate from? Because I think it’s like a wartime manufacturing thing. Is that right?

Len Greski
That’s right. It started in the 1940s as a practice at General Electric where they were trying to during the wartime period, reduce the cost of their products and services while maintaining an appropriate quality level. And value engineering is simply the analysis of a good or service in terms of the cost of its sub-components or inputs and the relative value that they contribute to the finished good.

And the idea is to find ways to deliver equivalent or improve value at the finished good level by reducing the cost of its sub-components. And so that approach, not only can it be used with parts and finished goods, like say like a vacuum cleaner or a jet engine, it can be used with a business process or a capability or a value stream to be able to understand the sources of cost and how they contribute to overall value that’s produced by your value stream, your business process, or a capability or an asset.

Stacy Gordon
So give me an example here. If I were at a company and I ran a business capability, what would be the first number you would ask me for?

Len Greski
Well, I would want to understand for labor and non-labor, what does it cost to operate that capability? And then I would want to know how it generates benefits so that I could per unit of time subtract benefits minus cost to see if the capability is above water, that it’s generating more value than cost.

Okay. More benefits than cost, because value is simply benefits minus cost per unit of time.

Stacy Gordon
Right. Well, you had a really great example in your article about this. Why don’t you tell me about the servers? I though that was pretty powerful.

Len Greski
Well, there are three or four different directions that I could take that question about servers, most of which require adult beverages to be fully appreciated.

Stacy Gordon
Oh, well, next time I’ll bring the Manhattans.

Len Greski
Since the sun’s not over the yard arm across most of the US, a straightforward answer is that the company in question had centralized its global infrastructure to the degree that the dozen or so business units consumed infrastructure from the centralized organization without having direct control over its costs. So infrastructure was treated as a vendor to the business units through a chargeback mechanism.

And since this particular company manufactured computer equipment, the infrastructure team could buy hardware at 30% of retail.

Stacy Gordon
Oh, wow. And

Len Greski
They also figured out that the more hardware they purchased, the more staff they needed operated. And so they over time realized that they could actually buy three times the amount of hardware you needed for a particular project or capability without raising suspicions in the finance group. And that allowed them to hire three times the amount of staff they needed to run operations.

And the chargeback mechanism obfuscated the accountability for the burgeoning infrastructure costs because the infrastructure department’s P&L statement was considered profitable because it had the authority to pass the costs along to the other departments without any accountability for the value or lack thereof that was actually being delivered. Now you’d wonder, wouldn’t the CIOs or the business units be concerned about that?

Well, the payoff to the CIOs is that the infrastructure was actually charged to the business P&L, not the business unit’s IT department.

Stacy Gordon
Oh.

Len Greski
So IT costs as a percentage of sales were frequently understated by 40 to 60%.

Stacy Gordon
Well, I hope that there’s a lot of people listening and understanding how to get around their overspend of next year or this year. That is interesting. Do you think today there are still a lot of companies that are typically doing this?

Len Greski
Well, it varies a lot. And in this particular company that they had really the supporting capabilities like HR and IT and finance would typically bill the other business units for their “services”. And so you had this very unusual culture where a department like IT could say, “I’m profitable because I’m bringing in more than chargebacks than I’m costing.”

But it was all interdepartmental chargebacks. There was no actual revenue being brought into the company. And in this particular company, it was so hard to untangle the web of chargebacks that they had to split the company. They had to basically divest it in order to force every dollar of cost to be assigned to one of the new businesses that was being spawned out of this company.

Stacy Gordon
Wow. So you literally had to divest to create a new company so you could practically start over because there was no way to untangle the web.

Len Greski
Yes.

Stacy Gordon
I mean, well, I always say Len, you have great stories. So why don’t you tell me the one about, which I really did love this one, about the argument between the two executives and who wanted to build the new… Again, you’ve already mentioned it, people love to build new capabilities, but understanding what that means and how that impacts the company as a whole, I though this was a really interesting story when you talk about value.

Len Greski
Well, if the podcast viewer’s interested in the long version of this story, I wrote an article called Getting Beyond Ourselves for Architecture and Governance Magazine back in 2022 that includes a more detailed fictionalized version where the names were changed to protect the guilty. The condensed version is that beyond the two senior executives, there are four actors in this drama, the e-commerce VP, the chief technology officer, the director of e-commerce customer service, and the business architect who did the value engineering assessment.

And through the agreement, it’s like for the chief technology officer, that debate went really well. He was successful because he avoided a $20 million unnecessary capital expense and the existing infrastructure worked as planned, as load increased, and over a three-year period, the systems worked as expected without having to make a big capital expenditure.

For the director of e-commerce customer service, her part of the story was that she had a significant problem with back orders for e-commerce orders. And what the value engineering analysis showed was that the most important thing to fix was her problem. So she got a good outcome because she finally got the attention she needed to fix her back order problem, generating millions of dollars a year in productivity gains and increasing customer satisfaction because you had a higher fulfillment rate.

Now the e-commerce business VP, things didn’t go so well because a 30% compound annual growth rate was baked into a sales plan.

Stacy Gordon
Oh no. And

Len Greski
Since his primary strategy for growing his business was, “Make it like WebKins, the virtual pet website for children,” he failed to achieve his growth target. And finally, for the business architect who conducted the analysis, it was a bit of a mixed bag. On one hand, the demand model that was used to assess the architecture scalability turned out to be accurate to within one half of 1%, 18 months in the future, which was a huge success and why the infrastructure didn’t fail under load, because the forecasting model was highly accurate.

But on the other hand, since the numbers didn’t support the e-commerce VP’s position and the architect was a direct report of the e-commerce VP, she was exiled to IT. And as an old friend of mine once said, “Righteousness is always an attractive option, but a true statement that sometimes takes years to play out, and that’s a story for another time.”

Stacy Gordon
Well, what I thought was interesting about this, I would’ve thought that most companies would have to hire a consulting firm to come in for value engineering. Can you oftentimes, do most larger companies have team members internally that can actually execute against an analysis like that without having third party objectivity?

Len Greski
Well, it’s many companies like say General Electric, they have a strong tradition of industrial engineering and have high quality finance departments and that those skills are readily available. I mean, they’re taught at universities, particularly either in industrial and systems engineering or master’s of business administration courses. I was first exposed to it in a purchasing management class, but because once you have a tool, the tool can be applied in multiple contexts and it’s knowing what tools to apply in what context that help you improve the operations, the value being generated by your business.

Stacy Gordon
Yeah. I just often wondered, it makes sense if you have a tool to be able to do that. It just was kind of interesting if you’d be swayed in one way or the other. I know in your article it talks about having to actually take something that people often say is intangible in cost and put a cost associated to it.

So I didn’t know if sometimes that’s maybe harder for someone to get to ground on that if you’re internal than it would be external.

Len Greski
Well, I mean, there is a big debate in both industry… It’s actually more challenging in economics and the social sciences of does everything that exists, can everything that exists be measured? And in the social sciences, in academia, you learn as an undergraduate that everything that exists can be measured, and then it’s just a question of how good you are at operationally measuring something.

So for example, many years ago, executives at the Museum of Science and Industry in Chicago wanted to figure out what the most popular exhibit was, and they sent out a survey and the survey came back with all kinds of bias in it that people were likely to mention snobbish exhibits that were scientifically complicated that were in the museum.

But if you actually looked at those locations in the museum, there were not many people that typically frequented those exhibits. So they asked some sociologists at the University of Chicago how they’d solve this problem, and the answer that came back was, “Let’s go talk to the maintenance engineers and figure out where they replace tiles, floor tiles throughout the building.”

And the winner was the baby chick exhibit. And so one could make an argument you can’t measure the most popular exhibit at the Museum of Science and Industry because people will lie to you on a survey. And the answer was actually you can measure it, but you just have to know what’s a better way to measure it.

Stacy Gordon
Sure, sure, sure, sure. Think outside of the box, I suppose. Kind of circling back on where we started, Len, we talked about AI is associated for being the cause of these labor cuts and the overage of operating expenses. When we look at value engineering, do you believe that if companies start focusing more on value engineering that this will stop the layoffs, or would you say it maybe just gives better math?

Len Greski
Well, as I think about that, I’ve got a couple of responses. First, better math is an obvious improvement over no math.

Stacy Gordon
I think that should be our headline. Better math is better than no math. Okay.

Len Greski
Next is that financial models help avoid making the wrong cuts or the wrong investments as the e-commerce story where $20 million in spend on a new e-commerce platform wouldn’t fix the back order problem in the supply chain, and its ability to create the growth promised by the e-commerce VP is dubious at best.

Now, a consistent theme in my coaching of senior executives is to start by fixing things that are under your control and to use the improved performance to develop influence with peers and stakeholders to change things that are at the boundaries of your responsibility. Because if you’re a good steward of your responsibilities, that’s going to create opportunities for influence and growth.

Because what happens often in productivity plays is people say, “Well, we’ll repurpose the capacity to something else,” and the capacity never gets repurposed to something that generates incremental value. So all you’ve done is invested in some productivity gain and increased your operating expense footprint. Right,

Stacy Gordon
Without solving the real problem.

Len Greski
Without monetizing the value, and that’s why I make such a big deal about monetizing the value of work in my most recent writing. But that fixing your own stuff, I like to call it get off the Pareto chart where Pareto chart is a common graphic that shows problems by frequency in categories of problems, how frequently they occur, relative to a cumulative percentage of all problems where the idea is like the Pareto principle that 20% of the things cause 80% of the problems.

If you can address those high frequency problems, you’ll get a lot of leverage in improving the performance of your business without having to immediately fix everything. When you focus on fixing your own stuff, other executives aren’t going to get angry because you’re admitting their performance improvements needed in your area. In fact, a lot of your colleagues will be happy to point out where your problems are.

If you can enlist them in helping you resolve the problems so that you’re not on the Pareto chart anymore, then you have the ability to influence other stakeholders.

Stacy Gordon
Sure. I think that’s a foundational message that most executives should live by get your house in order and by doing that, you’ll lead by example and you can be more impactful again at your boundaries or beyond your own job. So with that said, there’s got to be people listening today that have a budget problem right now that probably have taken away some tips from our conversation and are wondering where do they start?

How do they change the culture? Is it a cultural change or can it be a small change, kind of like what you were just talking about to really start to see the impact of a value engineering analysis within their own organization?

Len Greski
Well, culture is never the right place to start because culture is an effect, not a cause. As a sociologist, I define culture as a set of shared beliefs and practices that make up the informal rules for how people in a group interact. And because it’s informal, it’s something that takes shape over time and is caused by other things.

So I always start with there are two primary drivers of change in life. One is a compelling vision and the second is pain. 2000 years ago, a dude walked along the shores of the Sea of Galilee and said, “Come follow me.” And commercial fishermen dropped their nets and followed.

Why? He had a compelling vision. But the second driver, there are not many compelling visions. The second driver of change is pain.

People, when they feel enough pain, when the severity of their pain exceeds the fear of change, then they’re ready to have an admission of need. And today there are a few compelling visions. And in fact, the most commonly used business example of vision is John F. Kennedy’s goal to send a human being to the moon by the end of the 1960s.

Yet the reality is that was actually a response to pain, the pain of having been beaten space by the Sputnik satellite in October 1957. So if even the shining example of vision is actually a response to pain, then you’re more likely to make change happen when the pain is sufficient to create an admission of need.

And my service story about the obfuscated infrastructure costs is a really good example of pain driven change because My company was six months into planning for a corporate split and every dollar of cost in the parent company was going to be allocated to one of the two resulting companies. And the CIO of my division, the business unit that I was currently then working in, had been bragging about how his IT expense as a percentage of sales was only 1.5%.

And in a manufacturing business, that’s really good. But since the infrastructure was centralized across all the business units, those costs weren’t charged to his budget. And I was like the proverbial prophet out in the desert in a hair shirt yelling sinners repent, repeatedly warning my manager about an incoming tsunami of infrastructure costs.

And I was basically ignored until month 11 of a 12 month divestiture process where my boss’s boss was told to absorb hundreds of millions of dollars of infrastructure costs into his budget, moving his IT as a percent of sales metric from 1.5% to 3.9%. And all of a sudden the “time I had been wasting” on understanding our cost structure wasn’t so wasteful anymore.

And so there’s a piece of being able to see where the puck is going to know where the pain will be to be prepared with a solution. And all of a sudden, my math was instrumental in taking scores of millions of dollars of costs out of various cost structures in that business unit.

Stacy Gordon
Is there a part of… You do a great job all the time of breaking things down and this seems like it could be an extraordinary large undertaking, but can one person start small, like a cross-functional team of one small pain point to start exercising those muscles on how to think differently? I mean, could change start to happen as small as that?

Len Greski
Absolutely. I mean, in fact, at LiminalArc, when we work with our clients, we very frequently structure change, whether it’s in an organization or product development or building a business capability into vertical slices so that you can deliver meaningful units of value every 90 days, like I discuss in the new software economics where you want to be able to make work small enough so that you can deliver a unit of business value no less frequently than every 90 days, generate value and then use the value generated to fund the next thing you want to change.

It’s a good risk mitigation tool as well because you can only spend so much money in 90 days, whereas if you have a 24 month project plan and if you don’t know that what you’re doing isn’t going to work until month 23, you’ve got a lot more risk hanging out there.

Stacy Gordon
Right. Absolutely. Well, Len, this as always, I enjoy our conversations. You always bring enlightening information to the table to talk about, and you do it in a very funny way for me.

I want to call it again for those who haven’t had a chance. Len’s latest article was Why Value Engineering Beats AI-Driven Cost Cutting. Len, why don’t you tell, you’ve mentioned a few other things. I know that you love to share your thoughts and authorship.

Where are some places that people could maybe find you or connect with you or read some of your things?

Len Greski
Well, I have a number of articles and podcast videos on the LiminalArc website at liminalarc.co. I’ve been writing for Architecture and Governance Magazine at this point for about 17 years and got articles on architecture and governance.com, probably going back to at least 2019 because it previously was a dead tree version back when trade magazines still existed in a paper format.

Stacy Gordon
Some of us still like those. I still love the paper format. Well, Len, again, thank you so much. I really appreciate it.

Everybody, thanks so much for listening. It’s Stacy Gordon here with Future State Now. Give us a like, follow us so that make sure that you can hear our next podcast when it comes out soon. Thanks. Take care.