# Why AEC Firms Should Enforce Naming at the File-Save Dialog **By Dan Cumberland** · Published August 27, 2026 · Categories: AI Strategy ![Featured image for Why AEC Firms Should Enforce Naming at the File-Save Dialog](https://cms.dancumberlandlabs.com/uploads/featured_image_73be69d104.webp) Illustration: Dan Cumberland Labs with Gemini. > Your team has the naming convention. It's in the onboarding deck, probably pinned somewhere in the shared drive, maybe covered in a training session from six... Your team has the naming convention\. It's in the onboarding deck, probably pinned somewhere in the shared drive, maybe covered in a training session from six months ago\. And yet, somehow, someone just sent you a file called "Final\_v3\_USE\_THIS\_REAL\_ONE\.dwg\." Call it The 25% Compliance Problem\. According to Unanet's 2026 AEC Inspire Report, only 29% of AEC firms report high confidence in the data powering their AI tools[2](/blog/blog-the-25-compliance-problem#ref-2)— meaning roughly 3 in 4 firms are running operations on a foundation they don't trust\. This article explains why the naming convention you already have isn't working— and why that failure now carries a cost that didn't exist five years ago\. ## Why Naming Conventions Don't Stick Naming convention compliance fails because it's designed as a behavior\-change problem rather than a workflow\-design problem\. When the correct file name requires remembering a 10\-component schema under deadline pressure, non\-compliance is the path of least resistance\. That's a cognitive load problem\. CMAP/Atvero described it directly: "Manual naming conventions are unreliable in modern AEC environments, with file names such as 'Final\_v6' or 'Latest\_Approved' quickly creating confusion across teams and projects\."[8](/blog/blog-the-25-compliance-problem#ref-8) That sentence will be familiar to anyone who's managed a Common Data Environment \(CDE— the centralized document system where project files live\) for more than a month\. File names get creative, then they get unusable\. Three structural failure conditions drive this pattern: - **Cognitive load under delivery pressure\.** ISO 19650\-compliant naming schemas can require up to 10 data fields per file name\. Under deadline pressure, memorizing that schema is not the priority— getting the file sent is\. - **The memo rollout problem\.** Naming conventions announced via email and enforced by reminder lose traction within weeks\. Without infrastructure built into the workflow itself, behavior change collapses under delivery pressure\. That's how behavior change works\. - **No enforcement at the point of creation\.** The convention exists at the policy layer\. The file\-save dialog doesn't know anything about it\. That gap— between the policy document and the actual file\-save moment— is where compliance goes to die\. Practitioners consistently observe that without enforcement infrastructure, adoption typically falls below 20% within weeks of rollout\. But the mechanism matters more than the number\. For a deeper read on the architectural reasons compliance fails, [why naming convention architecture matters in AEC](/blog/why-do-we-need-architecture/) covers the underlying design logic\. The question worth asking is why teams stop using a convention that's correct, and why that failure is structural\. ## What the Compliance Gap Costs Bad data cost the global construction industry an estimated $1\.8 trillion in 2020, according to a 2021 Autodesk and FMI Consulting study of nearly 4,000 construction professionals[1](/blog/blog-the-25-compliance-problem#ref-1)\. Naming and metadata compliance is one consistent contributor to that figure among several\. And its daily cost shows up in every employee who spends time searching for a file that should have been findable in seconds\. Research from McKinsey and IDC consistently shows that knowledge workers spend 1\.8 to 2\.5 hours per day— up to 30% of the workday— searching for information they can't readily find[3](/blog/blog-the-25-compliance-problem#ref-3)[4](/blog/blog-the-25-compliance-problem#ref-4)\. Apply that to an AEC project team: if 10 people on a project spend 20 extra minutes daily tracking down poorly\-named files, that's more than 3 hours of lost labor per day, per project\. Across a portfolio, the math gets ugly fast\. ```html-table
Cost CategoryMetricSource
Global construction losses from bad data$1.8 trillion (2020)Autodesk/FMI 2021
Per $1B contractor$165M total bad data cost, including $1.7M in reworkAutodesk/FMI 2021
Avoidable rework linked to bad data14% of rework costsAutodesk/FMI 2021
Daily search time per employee1.8–2.5 hours (23–30% of workday)McKinsey 2012, IDC
``` And 30% of construction professionals in that same study reported that more than half their project data was inaccurate, incomplete, or inconsistent[1](/blog/blog-the-25-compliance-problem#ref-1)\. That's the majority of the data on the majority of projects failing a basic reliability test\. Those costs have always been real\. But in 2026, a new cost has been added— one that makes compliance failure a competitive liability, not just a productivity drag\. ## What AI Adoption Actually Demands from Your Data Seventy\-five percent of AEC firms now use AI tools— up roughly 20 percentage points in a single year, according to Unanet's 2026 AEC Inspire Report[2](/blog/blog-the-25-compliance-problem#ref-2)\. But that same report found only 29% of those firms have high confidence in the data those AI tools are running on[2](/blog/blog-the-25-compliance-problem#ref-2)\. Most AEC firms are using AI on a foundation they don't trust\. According to AEC Foundry[5](/blog/blog-the-25-compliance-problem#ref-5), over 80% of AI project time is spent on data engineering— collecting, cleaning, and organizing data\. In practical terms: a firm with poor naming compliance is buying data remediation before it buys any AI capability\. The MIT NANDA Initiative estimates that up to 85% of AI projects fail due to poor data quality[6](/blog/blog-the-25-compliance-problem#ref-6)\. And 45% of construction firms still lack a formal data strategy entirely[1](/blog/blog-the-25-compliance-problem#ref-1)— meaning they don't have an AI strategy either, whether they realize it or not\. AEC data is uniquely fragmented across BIM models, CAD drawings, technical PDFs, schedules, and spreadsheets— five different systems that don't share a naming logic\. Unified AI retrieval across all of them requires consistent metadata throughout\. Without it, cross\-system retrieval fails regardless of how capable the AI tool is: ask a well\-built AI assistant to reconcile the CAD revision with the matching spec PDF, and it fails because nothing links the two\. Three AI readiness consequences that follow directly from naming non\-compliance: - AI tools surface wrong or outdated documents, eroding user trust and reducing adoption - Data preparation consumes the AI project budget before capabilities are built - Cross\-system retrieval across BIM, CAD, and PDF systems fails without consistent metadata Understanding [AI governance strategy](/blog/ai-governance-strategy) is part of building a firm that can use AI effectively\. But governance without clean data infrastructure is a policy document sitting on top of a broken foundation\. Before committing to a build, it's worth understanding the [hidden costs of AI projects](/blog/hidden-costs-ai-projects)— data remediation is often the largest one, and it's rarely in the initial budget\. There's another cost that doesn't show up in any of these studies— one that becomes permanent\. ## The Institutional Knowledge Problem According to IDC estimates[7](/blog/blog-the-25-compliance-problem#ref-7), 85% of enterprise data is unstructured— sitting in formats and file systems that aren't organized for systematic retrieval\. In AEC, that means project knowledge is spread across inconsistently named drawings, outdated spec versions with no clear revision history, and email chains that never made it into the CDE\. And 80% of organizations fail to classify data for compliance at all[7](/blog/blog-the-25-compliance-problem#ref-7)\. When a senior project manager retires, the files they navigated by memory become opaque to everyone else\. The naming convention— had it been followed— would have been the system that made that knowledge accessible\. Without it, the knowledge leaves when the person does\. > "A file named 'Client\_Final\_FINAL\_v6\_SendThis\.dwg' encodes institutional knowledge in a way only its creator understands\." Building good data infrastructure is like building an iceberg from the bottom up— the compliance work is invisible until AI shows up and demands a foundation\. And when it does, firms find out whether they've built one\. The good news: this is a design problem\. Design problems have design solutions\. ## The Fix Is Structural, Not Behavioral The naming convention compliance problem gets solved by embedding compliance into the creation workflow— so the correct file name is the only available option when a file is saved, rather than one of many choices made under deadline pressure\. Just because it's easy to send a policy doesn't mean it produces good results\. Designing compliance into workflows takes real upfront investment— but it pays returns across every project that follows\. The mechanisms exist: CDE configurations, automated naming plugins for Revit and Autodesk Construction Cloud \(or BIM 360, if your firm hasn't migrated yet\), document control templates that enforce structure at the point of creation\. This is infrastructure work\. Compliance embedded in workflow beats compliance enforced by policy\. Every time\. If you're figuring out where to start, [AI implementation services](/services/ai-implementation/) can help map the right approach to your existing systems— without adding another policy layer that won't stick\. ## FAQ ### What is the 25% compliance problem in AEC? Only about 29% of AEC firms report high confidence in the data powering their AI tools, according to Unanet's 2026 AEC Inspire Report[2](/blog/blog-the-25-compliance-problem#ref-2)\. That means roughly 3 in 4 firms are running operations— and now AI systems— on a data foundation they don't fully trust\. The "25% compliance problem" refers to this gap: most firms have naming conventions, but most see those conventions break down in practice\. ### Why do AEC naming conventions fail? Most naming convention rollouts rely on behavior change \(send a memo, hold a training\) rather than workflow design\. Under deadline pressure, behavior change loses\. When the correct file name requires remembering a 10\-component schema, non\-compliance becomes the path of least resistance\. ### How much does bad data cost a construction firm? A 2021 study by Autodesk and FMI Consulting estimated that bad data cost the global construction industry $1\.8 trillion in 2020[1](/blog/blog-the-25-compliance-problem#ref-1)— the same figure driving the AI\-readiness gap covered above, here scaled down to the individual\-contractor level: a $1 billion contractor absorbs an estimated $165 million in bad\-data costs, including $1\.7 million in avoidable rework\. ### What does naming compliance have to do with AI? Over 80% of AI project time is spent collecting, cleaning, and organizing data, according to AEC Foundry[5](/blog/blog-the-25-compliance-problem#ref-5)\. In firms with poor naming compliance, that time— and budget— is consumed before any AI capability is built\. AI systems also require consistent metadata to retrieve and connect information across BIM models, CAD drawings, PDFs, and schedules— without it, cross\-system retrieval fails\. ### How many employees lose time searching for files? Research from McKinsey and IDC shows knowledge workers spend 1\.8 to 2\.5 hours per day— roughly 23–30% of the workday— searching for information they can't readily find[3](/blog/blog-the-25-compliance-problem#ref-3)[4](/blog/blog-the-25-compliance-problem#ref-4)\. In AEC project environments, where the correct document version can determine whether a project runs on\-spec, that search cost compounds across every project and every team member\. Workers require up to 8 searches to locate the correct document[4](/blog/blog-the-25-compliance-problem#ref-4)\. ## Where Compliance Actually Starts The naming convention compliance problem is solvable\. Firms that fix their data infrastructure now will build AI capabilities faster than firms that don't; in a market where AI adoption is accelerating, that gap compounds every quarter it goes unaddressed\. How fast depends on where you're starting from\. The fix starts at the file\-save dialog your team already uses today— which is where compliance either holds or stops\. For a practical starting point on [building an AI\-ready culture](/building-ai-culture/), begin with the data foundation your AI will depend on\. ## References 1. Autodesk & FMI Consulting, "New Report Reveals Data Strategy is a Key Advantage in Construction" \(2021\)— [https://www\.autodesk\.com/blogs/construction/construction\-data\-strategy\-report/](https://www.autodesk.com/blogs/construction/construction-data-strategy-report/) 2. Unanet, "2026 AEC Inspire Report: AI Adoption Surge While Data Confidence Lags" \(2026\)— [https://unanet\.com/news/unanet\-releases\-2026\-aec\-inspire\-report\-revealing\-ai\-adoption\-surge\-while\-data\-confidence\-lags](https://unanet.com/news/unanet-releases-2026-aec-inspire-report-revealing-ai-adoption-surge-while-data-confidence-lags) 3. McKinsey Global Institute, "The Social Economy: Unlocking Value and Productivity Through Social Technologies" \(2012\)— [https://cottrillresearch\.com/various\-survey\-statistics\-workers\-spend\-too\-much\-time\-searching\-for\-information/](https://cottrillresearch.com/various-survey-statistics-workers-spend-too-much-time-searching-for-information/) 4. IDC Research, "Various Survey Statistics: Workers Spend Too Much Time Searching for Information" \(2023\)— [https://cottrillresearch\.com/various\-survey\-statistics\-workers\-spend\-too\-much\-time\-searching\-for\-information/](https://cottrillresearch.com/various-survey-statistics-workers-spend-too-much-time-searching-for-information/) 5. AEC Foundry, "Making AEC Data Work for AI: A Practical Playbook for Search, Agentic Workflows, and Governance" \(2025\)— [https://www\.aecfoundry\.com/blog/making\-aec\-data\-work\-for\-ai\-a\-practical\-playbook\-for\-search\-agentic\-workflows\-and\-governance](https://www.aecfoundry.com/blog/making-aec-data-work-for-ai-a-practical-playbook-for-search-agentic-workflows-and-governance) 6. MIT NANDA Initiative, "AI Construction Statistics for 2026" \(2026, via Bridgit\)— [https://gobridgit\.com/blog/ai\-construction\-statistics/](https://gobridgit.com/blog/ai-construction-statistics/) 7. IDC / WorldMetrics, "Dark Data Statistics 2026" \(2026\)— [https://worldmetrics\.org/dark\-data\-statistics/](https://worldmetrics.org/dark-data-statistics/) 8. CMAP/Atvero, "Guaranteeing Strict File Naming to Become ISO 19650 Compliant" \(2024\)— [https://www\.cmap\.io/blog/guaranteeing\-strict\-file\-naming\-to\-become\-iso\-19650\-compliant](https://www.cmap.io/blog/guaranteeing-strict-file-naming-to-become-iso-19650-compliant) --- **How we made this article:** We use AI in our research and writing so our small team can share more of what we learn. We verify the sources and take responsibility for every article we publish. [Read how we use AI.](https://dancumberlandlabs.com/how-we-use-ai/) --- ## About the author **Dan Cumberland** — Founder, Dan Cumberland Labs Dan Cumberland helps engineering and construction firms see where they stand with AI and decide what to build first. He created Pacemark, the AI maturity model behind that work, from research on more than 300 companies. - Take the assessment: https://pacemark.ai/signal/assessment/?track=aec&utm_source=dcl-site&utm_medium=link&utm_campaign=pacemark-assessment - Book a call: https://book.dancumberland.com/ai-strategy --- Source: https://dancumberlandlabs.com/blog/the-25-compliance-problem/