Not All Tech Debt Is Worth Paying Down

The Paradox: More Tech, Same Productivity

The average AEC firm now manages 86.8 software applications, yet construction productivity has grown only 10% globally over the last 22 years. The modernization story isn't producing the returns the industry was promised. And the right response to bad tech construction outcomes isn't more tech. It's better tech selection.

The data:

• AEC firms manage an average of 86.8 software applications per firm⁶
• Construction productivity grew just 10% globally from 2000 to 2022⁴, one-fifth the rate of the overall economy
• 95% of enterprise GenAI pilots produce no measurable P&L impact¹
• Just 16% of digital transformations deliver sustainable performance improvements⁵

A senior partner at a $40M AEC firm sits through another vendor pitch this week. The slide deck promises transformation. The pricing assumes urgency. The data above tells a different story. The industry has been buying these pitches for two decades, and productivity hasn't moved.

The modernization narrative tells AEC leaders to keep upgrading. The data tells them to keep deciding.

This article is for the partner, principal, COO, or CFO of a $20M–$100M AEC firm who suspects something is off in the stack but doesn't have the framework to name it. Most coverage in this space pushes one direction: modernize faster. This piece pushes a different one: choose better. More tools without strategic discipline produce more tech debt, not more productivity.

Before deciding which systems to keep, leaders need a sharper definition of what "tech debt" actually means inside an AEC firm. The term gets used loosely, and that looseness is part of the problem.

What "Tech Debt" Actually Means Inside an AEC Firm

In construction, technical debt is the accumulated cost of past technology decisions that limit future flexibility: legacy ERP customizations, unintegrated point solutions, accumulated SaaS subscriptions, and AI pilots that never reached production. CFMA research⁷ found 75% of construction tech debt originates in legacy software, sometimes from systems built in the late 1980s and 1990s.

Construction-specific tech debt usually shows up in four places:

• Legacy ERPs still in use 20+ years after deployment, customized in ways that block upgrades⁸
• Excel as integration glue between systems that don't talk to each other, with each manual export creating new error surface
• Redundant point solutions acquired by different departments solving similar problems independently
• Abandoned subscriptions and stalled AI pilots still on the company card but no longer producing value

McKinsey estimates tech debt accounts for 20–40% of an organization's entire technology estate value³. That's not a software developer problem. It's a leadership problem disguised as an IT problem.

The distinction worth drawing is between visible debt and invisible debt. Visible debt is loud: the system everyone complains about, the workaround that costs hours each week. Invisible debt is silent: the customization no one remembers, the integration that almost works, the subscription paid out of habit. Most firms find their visible debt. But most miss the invisible kind.

Every AEC firm has tech debt. The question is whether the debt is strategic or accidental.

Martin Fowler's Tech Debt Quadrant, Translated for AEC

Martin Fowler's Technical Debt Quadrant² categorizes debt along two dimensions: deliberate vs. inadvertent and prudent vs. reckless. Applied to AEC operations, the framework reveals a counterintuitive truth. Prudent deliberate debt is a strategic asset worth keeping, not a failure to fix.

Fowler published the framework in 2009 for software engineering. The translation to AEC operations is straightforward when each quadrant gets a concrete construction example.

Prudent deliberate tech debt is a strategic mortgage. In practical terms: it's a conscious shortcut taken to gain something more valuable. Replacing it isn't paying down debt. It's wasted spend.

Reckless debt accumulates when departments buy software independently. Prudent debt accumulates when leadership chooses to. The difference matters because most AEC firms treat all debt the same: a problem to fix on the next budget cycle.

The leadership move is recognizing which quadrant each system in the stack actually sits in. Most $20M–$100M firms have never produced this inventory. Without it, every modernization conversation defaults to the same pattern: vendor pitch arrives, urgency gets manufactured, capital gets allocated, and the next pitch starts the cycle again.

This isn't a critique of vendors doing their jobs. It's a critique of leadership skipping its own. An AI decision framework for founders lives upstream of every individual purchase decision, and the same principle applies inside an AEC firm.

Naming each system's quadrant is the diagnosis. The next step is the prescription, and Gartner has been quietly providing one for years.

Why Most Construction Modernizations Fail

Most construction digital transformations don't deliver. McKinsey found just 16% of digital transformation respondents reported sustainable performance improvements⁵. The average company spent $1.9 million on GenAI in 2024, with less than 30% of CEOs satisfied with returns¹.

The pattern behind the failures is consistent:

• Technology-first thinking. The tool gets selected before the operations problem is named. Implementation reveals the workflow it was supposed to fix wasn't actually broken.
• Weak change management. The platform launches but adoption never sticks. Field teams keep their workarounds.
• Integration sprawl. The new tool joins the stack instead of replacing anything. More apps, more accounts, more places for data to live.

McKinsey's framing of the pattern: "few engineering and construction companies have captured the full benefit of digital"⁵. That isn't a critique of technology. It's a critique of selection discipline. The technology works. The decision process around it doesn't.

This is where the hidden costs of AI projects compound. Every failed modernization adds to the stack, drains capital that could have funded the next disciplined investment, and erodes leadership credibility for the next genuinely valuable initiative.

The 16% who succeed aren't using better technology than the 84% who don't. They're using better decision frameworks.

If the failure pattern is undisciplined selection, the fix is a decision framework leaders can run on a Tuesday morning.

The TIME Framework: A Tech Construction Decision Tool for Each System

Gartner's TIME framework¹¹ gives AEC leaders four decisions for each system in their stack: Tolerate, Invest, Migrate, or Eliminate. Combined with total cost of ownership analysis, it turns "should we modernize?" into a series of system-by-system answers leaders can defend in a partners meeting.

Every system in your stack belongs in one of four categories. The discipline is naming which one and acting on it.

The mapping back to Fowler is the part most leaders haven't seen drawn. Tolerate ≈ Prudent debt: the discipline of leaving working systems alone. Eliminate ≈ Reckless debt: the discipline of retiring what shouldn't have been bought. Migrate ≈ debt that's become genuinely costly.

The TIME vocabulary works in the rooms where decisions actually get made. A CFO asking "what's our IT roadmap?" gets a different answer when leadership can hand back a list with each system tagged and TCO numbers attached. A partner asking "why are we still using that old job-cost system?" gets a defensible answer when the response is "we tagged it Tolerate last quarter, here's the TCO comparison against the alternatives."

Once each system has a tag, each tag has to be defended. That's the part most internal IT roadmaps miss.

The hardest of the four to defend in a leadership meeting is also the most important: Tolerate. That deserves its own discussion.

The Discipline of "Good Enough"

Working systems with documented limitations are often a better strategic asset than new systems with unknown ones. The construction industry's 1.4% IT spend, lowest of major industries⁹, isn't insufficient when 86.8 apps per firm⁶ shows the issue is allocation discipline, not budget size.

Pair the 1.4% revenue figure (against a 3.6–5% cross-industry average) with the 86.8-apps-per-firm benchmark and the picture changes. The industry isn't spending too little. It's spending without selection.

"The AEC industry is drowning in software."⁶— Randall Stevens, AVAIL CEO

Drowning in software is the right diagnosis. Flying out of it starts with deciding which apps stay, which go, and why.

Three criteria justify the Tolerate decision:

• The system meets functional requirements teams actually use
• Replacement total cost of ownership exceeds documented benefit
• Staff can use the system effectively today, and migration would create real disruption

Vendor pitches deserve the same skepticism as any major capital decision. Vendors profit from sales. Trade press is often vendor-funded. That doesn't make modernization wrong. It makes "modernize because the pitch sounds urgent" wrong.

You can't read the label from inside the bottle. Leaders running their stacks every day are too close to see them clearly. An external framework, Fowler's quadrant or Gartner's TIME, is the label viewed from outside.

Permission to keep what works isn't laziness. It's the discipline most AEC firms haven't applied yet.

AI Is the Next Tech Debt Accumulator

AI is the next category of tech debt AEC firms will accumulate without strategic selection. MIT's 2025 research¹ found 95% of enterprise GenAI pilots produced no measurable P&L impact, while specialized vendor partnerships succeeded at 67% versus 33% for internal builds.

Construction sentiment is moving fast. 86% of large contractors and 69% of small/mid-sized firms believe AI will provide competitive advantage¹⁰. But sentiment isn't strategy. Most firms moving forward on that belief won't end up in the 5% who succeed.

The disciplined response is a three-question filter for every AI pilot:

• Is the operations problem named first? AI without a named problem is reckless deliberate debt.
• Is there a vendor with a track record? The MIT data is stark: vendor partnerships succeed twice as often as internal builds.
• Is there a kill criterion? If the pilot can't define what "stop running this" looks like in 90 days, it will run forever and become invisible debt.

Measuring AI success is the practice that separates the 5% from the 95%. Same with the AI consultant vs in-house build decision. The data on 67% vendor success vs 33% internal builds isn't ambiguous.

Not anti-AI. Anti-undisciplined-AI. AI doesn't fail because the technology is wrong. It fails because the selection isn't disciplined.

All of this collapses to a single question every AEC leader can answer this quarter: which quadrant is each system in?

How to Run a Tech Construction Audit This Quarter

A useful tech debt audit takes one afternoon and a list of every software subscription on the company card. Apply Fowler's quadrant to each system, then assign a TIME decision. The result is a stack-level inventory most $20M–$100M AEC firms have never produced.

The five-step audit:

1. List every software system on the company card. Most firms underestimate the count when asked. Pull credit card statements and accounts payable for the last 12 months. The 86.8-apps-per-firm benchmark⁶ is a useful sanity check.
2. Tag each with Fowler's quadrant. Prudent/Deliberate, Prudent/Inadvertent, Reckless/Deliberate, Reckless/Inadvertent. Be honest. This is where most lists get sanitized.
3. Assign a TIME decision to each system: Tolerate, Invest, Migrate, or Eliminate¹¹. Force the choice. No "TBD."
4. Total cost the Migrate and Eliminate items. This becomes the modernization budget for the year. Cancellations free capital. Migrations spend it. The math has to balance.
5. Defend the Tolerate decisions to the partners. This is where the discipline gets tested. If "Tolerate" can't survive a leadership conversation, it isn't actually Tolerate. It's avoidance.

That's a one-afternoon engagement, not a six-month consulting project. The output document alone reframes every modernization conversation that follows.

Most firms have never done this. That's the opportunity.

Frequently Asked Questions

What is tech debt in construction?

Tech debt in construction is the accumulated cost of past technology decisions that limit future flexibility: legacy ERP customizations, unintegrated point solutions, and accumulated subscriptions. CFMA found 75% of construction tech debt originates in legacy software⁷, sometimes from systems built in the late 1980s and 1990s.

Should construction firms replace all old technology?

No. Martin Fowler's Technical Debt Quadrant² distinguishes prudent deliberate debt (strategic, worth keeping) from reckless debt (worth replacing). Gartner's TIME framework¹¹ gives four options for each system: Tolerate, Invest, Migrate, or Eliminate.

Why do most construction digital transformations fail?

McKinsey found just 16% of digital transformations deliver sustainable performance improvements⁵. Common causes include technology-first thinking, weak change management, and integration sprawl: failures of selection discipline, not technology capability.

What is the average tech stack size for an AEC firm?

According to AVAIL's 2025 AEC Technology Stack Survey⁶, the average AEC firm uses 86.8 software applications across 20 categories, drawn from 410 distinct applications surveyed.

Are AI pilots successful in construction?

The MIT NANDA 2025 report¹ found 95% of enterprise GenAI pilots produce no measurable P&L impact. Specialized vendor partnerships succeed at 67% versus 33% for internal builds.

How much should a construction firm spend on IT?

Construction averages 1.4% of revenue on IT, the lowest of major industries⁹ (cross-industry average is 3.6–5%). The strategic question is allocation discipline, not just total spend.

The Smart Move

Not all tech debt is worth paying down. And the leaders who recognize that will allocate their next dollar of IT budget more strategically than the ones who keep saying yes to every modernization pitch.

The smart move isn't "rip and replace." It's "name the quadrant, assign the decision, defend the choice." This is harder than blanket yes or blanket no. That's why most firms don't do it.

Both are true: modernization pressure is real, and blanket modernization fails. The discipline lives in deciding system by system.

If walking your team through this framework— Fowler's quadrant, Gartner's TIME, applied to your stack with TCO numbers attached— would be useful, that's exactly the AI strategy work for AEC firms we do. Not a six-month engagement. A working session that produces the inventory document most firms have never had.

References

1. MIT Project NANDA / Fortune, "MIT report: 95% of generative AI pilots at companies are failing" (2025) — https://fortune.com/2025/08/18/mit-report-95-percent-generative-ai-pilots-at-companies-failing-cfo/
2. Martin Fowler, "Technical Debt Quadrant" (2009) — https://martinfowler.com/bliki/TechnicalDebtQuadrant.html
3. McKinsey & Company, "Tech debt: Reclaiming tech equity" (2020) — https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/tech-debt-reclaiming-tech-equity
4. McKinsey Global Institute, "Delivering on construction productivity is no longer optional" (2024) — https://www.mckinsey.com/capabilities/operations/our-insights/delivering-on-construction-productivity-is-no-longer-optional
5. McKinsey & Company, "Decoding digital transformation in construction" (2016) — https://www.mckinsey.com/capabilities/operations/our-insights/decoding-digital-transformation-in-construction
6. AVAIL, "AVAIL Releases 2025 AEC Technology Stack Survey Results" (2026) — https://blog.getavail.com/tech-stack-survey-pr
7. Construction Financial Management Association (CFMA), "Reducing Technical Debt: Strategies for Eliminating & Avoiding Limitations With Legacy Systems" (2024) — https://cfma.org/articles/reducing-technical-debt-strategies-for-eliminating-and-avoiding-limitations-with-legacy-systems
8. Construction Financial Management Association (CFMA), "Navigating the Evolving Landscape of Construction ERP" (2024) — https://cfma.org/articles/navigating-the-evolving-landscape-of-construction-erp
9. Avasant / Computer Economics, "IT Spending as a Percentage of Revenue by Industry, Company Size, and Region" (2024) — https://avasant.com/report/it-spending-as-a-percentage-of-revenue-by-industry-company-size-and-region/
10. Construction Dive (covering Dodge Construction Network/CMiC research), "AI nears 'tipping point' in construction as contractors pilot tech: survey" (2025) — https://www.constructiondive.com/news/builders-ai-transform-businesses-survey/807555/
11. Gartner, "Reduce and Manage Technical Debt" (2024) — https://www.gartner.com/en/infrastructure-and-it-operations-leaders/topics/technical-debt