The request that finally got someone's attention was not complicated. A commercial insurance carrier we worked with wanted to offer a usage-based product to its mid-market fleet customers. Telematics data was already flowing in from the vehicles. The actuaries had the model. Sales had three named accounts waiting to sign.

The IT estimate came back at eleven months.

Not eleven months to build the pricing engine. That part was maybe six weeks. Eleven months because the rating logic lived inside a policy administration system written in the late 1990s, the system had no API surface, the only path to it ran through a nightly batch file, and the three people who understood the batch schedule were the same three people already committed to a regulatory change that could not slip.

While the estimate was being socialized, a competitor launched something similar. Not a startup. Another carrier of similar size, in the same regulatory environment, with the same actuarial constraints. They shipped in seven weeks.

The difference was not talent, budget, or ambition. The CIO had a bigger team and a bigger budget than the competitor did. The difference was that one company could change its systems and the other could only maintain them.

What was actually in the way

We spent three weeks on the assessment, and none of what we found was exotic.

Fourteen point-to-point integrations connected the policy system to billing, claims, print, the broker portal, and four external data providers. Every one of them had been built by a different team in a different decade under a different deadline. There was no interface layer, so any change to the policy system meant regression testing all fourteen.

The release train ran quarterly, because that was the only cadence at which the regression suite could be completed by hand. Which meant that a two-day code change waited an average of six weeks for a window, and any defect found in the window pushed the whole train.

Roughly 40% of the application spend was going to keeping things running, and that number understated it, because the change budget was mostly regulatory work the business had no choice about. Discretionary spend, the money actually available to build something a customer would notice, was under 15% of the total.

And the data the actuaries wanted was technically available and practically unreachable. Customer records were duplicated across four systems with no common identifier. Answering "what is this account's total relationship with us" was a two-week manual exercise.

None of this was a surprise to anyone inside the company. Every one of these facts had appeared in a slide deck at some point over the previous five years. What had never happened was anyone putting a number on what it cost.

The number nobody had calculated

The industry data says this pattern is close to universal.

up to 70%

of technology budgets go to run-the-business work at large organizations, per McKinsey

30%

of IT budget goes to servicing technical debt on average, per Protiviti's survey of 1,000+ technology executives

$370M

lost annually by the average global enterprise to inefficient modernization, per Pegasystems and Savanta

In that same Protiviti survey, close to 70% of executives said technical debt materially limits their ability to innovate. NTT Data puts the figure higher still, with 80% of organizations reporting that outdated technology is constraining their innovation efforts. These are not IT complaints. They are leaders naming their own architecture as the thing standing between them and the plan they committed to.

For the carrier, the calculation we ran was simple and uncomfortable. Not the cost of the legacy system, which they knew. The cost of the decisions they had declined to make because of it. Three product launches deferred over four years, one distribution partnership lost during due diligence when the partner reviewed integration timelines, and a book of business their own sales team had stopped pitching because the operational drag made it unprofitable to service.

That number was multiples of the modernization estimate everyone had been avoiding.

Where the Harvard research changes the conversation

The most useful framing we have found for this comes out of Harvard Business Review Press. In Future Ready: The Four Pathways to Capturing Digital Value, Stephanie Woerner, Peter Weill, and Ina Sebastian of MIT's Center for Information Systems Research published a study of 1,311 global firms built on more than five years of research and surveys of over 2,000 respondents.

Two findings from that work matter for any board looking at a modernization request.

The first is the performance gap. Only 22% of firms in the study qualified as future ready, meaning they had done the transformation work that lets them innovate and improve customer experience while simultaneously reducing cost.

22%

of 1,311 global firms qualified as future ready

+17.3 pts

revenue growth above industry average for those firms

+14.0 pts

net margin above industry average for those firms

Related CISR work found that companies more than halfway through their transformation had 32% higher net margins and 67% higher revenue growth than companies less than halfway.

The second finding is the diagnosis, and it is the one that lands with executives. The researchers place companies on a grid, and the starting quadrant for most established enterprises is the one they call Silos and Spaghetti. Silos are the business units and product lines that each own their own view of the customer. Spaghetti is the tangle of connections holding them together. Both were rational when they were built. Together they make the company structurally slow.

The carrier did not have a technology problem in the sense of old software. It had a structural problem where the cost of change had risen high enough that the business had quietly stopped asking for change. That condition does not appear in any operating report. Demand disappears before it reaches IT.

The research also makes clear that there are multiple valid routes forward, and picking the wrong one for your starting position is a common and expensive error. Fixing operations first, fixing customer experience first, alternating between the two, or standing up a separate new business are all defensible. What is not defensible is choosing by default.

What this means if it sounds familiar

Two things are different about this moment, and both raise the cost of waiting.

The first is AI. Every agentic use case being pitched to your board right now depends on two things your legacy estate probably cannot provide: programmatic access to systems of record, and data clean enough for a machine to act on without a human checking. Companies are currently discovering that their AI pilots are failing for reasons that have nothing to do with the models. The modernization work is the AI work. Skipping it does not save money, it moves the failure downstream and makes it more expensive.

The second is that the gap compounds. The Harvard research is a snapshot of a widening distribution, not a static one. Firms that can change accumulate advantages that make the next change cheaper. Firms that cannot accumulate constraints that make the next change harder. There is no equilibrium in the middle.

If you want a starting point that does not require a program budget, ask three questions at your next operating review.

What is the average elapsed time from a business decision to that change being live for customers, and how does it compare to two years ago?
What share of technology spend is discretionary after regulatory, security, and keeping the lights on?
Which requests has the business stopped making because it already knows the answer?

The third question is the one that matters. When your commercial team stops asking, the constraint has moved from your systems into your strategy, and it will not show up in any report until a competitor makes it visible.

This story is a composite drawn from modernization engagements in insurance and financial services. Details have been changed. The pattern has not.

Referenced research: Stephanie L. Woerner, Peter Weill, and Ina M. Sebastian, Future Ready: The Four Pathways to Capturing Digital Value, Harvard Business Review Press, 2022, based on MIT CISR research covering 1,311 global firms; MIT CISR research briefings on transformation progress and financial performance; McKinsey research on run-the-business technology spending; Protiviti global technology executive survey (n=1,000+); NTT Data research on outdated technology and innovation; Pegasystems and Savanta legacy modernization research (n=500+).