Your Data Is Digital. That Was the Easy Part.
Ask most engineering firm leaders whether their firm has a data problem and you are likely to get a confident no. The drawings and models are digital, and people can find what they need. By the standards the industry set for itself fifteen years ago, that is a finished project.
The ACEC Research Institute’s new study, From Project Data to Infrastructure Intelligence, developed in partnership with Bentley Systems, suggests the standard has moved. Based on a survey of 251 member firms and interviews with firms and public owners, it finds an industry that has largely solved digitization and now faces a harder question: can that data still be used when the system, the organization, the project, or the purpose changes?
For a lot of firms, the honest answer is not yet.
The foundations are real. The connections are thinner.
The basics are largely in place: 88 percent of firms report that employees can readily access the project data they need, 78 percent use consistent data structures and naming conventions, and 76 percent say employees can discover relevant data from previous projects.
The trouble starts when data has to move or connect. Sixty-five percent say relationships and dependencies among project data are preserved. Sixty-one percent say data can move between major software applications without substantial reconstruction or re-entry. And only 44 percent say a change in one dataset or model automatically identifies or triggers impacts elsewhere.
The pattern is clear: firms are strongest at making data accessible and organized, weaker at making it move, and weakest where connectivity requires systems to recognize relationships and act on them. The underlying technologies tell the same story: only 23 percent of firms use APIs to connect systems, 21 percent use automated data integration, and 16 percent use open data or exchange standards.
Firms know where this needs to go. Seventy-four percent believe open standards and interoperability will be essential to achieving the full value of connected data, yet only 16 percent use them today. The conviction has arrived ahead of the implementation.
The cost is absorbed, which is why it is easy to miss
Disconnected data rarely produces dramatic project failures. It produces friction, distributed across hundreds of ordinary tasks: a junior engineer rebuilding a model that already existed, a QA/QC cycle that takes longer than it should, a handoff that requires a phone call to interpret.
Thirty-nine percent of respondents say searching for, validating, or verifying project data moderately reduces employee productivity. Fifty-five percent sometimes have to substantially rework data received from clients, partners, or contractors; 19 percent say it happens often. Moving data between applications is the most-cited difficulty, named by 39 percent.
None of that shows up on a project post-mortem, but it adds up, and the firms with the weakest data environments feel it hardest: 23 percent of firms rating their capabilities Fair or Poor report a significant or extreme productivity impact, compared with 6 percent of firms at the top of the scale.
AI is not a reason to wait. It is the reason the clock is running.
AI is already exposing the gaps. More than half of firms (53 percent) say poorly structured or fragmented data limits their ability to use AI effectively. Sixty-five percent say AI has made improving their data environment more important, and 73 percent say connected data will be essential to realizing AI’s full value.
That does not mean a firm must perfect its data environment before deploying AI. The interviews pointed toward parallel development: using AI while improving structure, quality, and governance.
But there is a reason AI makes data quality newly visible. An experienced engineer knows which of three similarly named files is the current one. That tacit knowledge has been quietly compensating for data environments that were never as clean as the org chart suggested. Automated systems do not have it. AI uses whatever data is readily available, whether or not it is the right data.
The hardest barriers are not technical
The most common constraints are insufficient staff or resources dedicated to connected data work (44 percent), lack of time and competing project priorities (43 percent), and insufficient internal skills (36 percent). Technology constraints such as legacy systems and proprietary formats rank lower. About nine in ten firms identify at least one barrier, and the typical firm identifies six.
Governance shows the same split between the basics and the harder work. Seventy-one percent of firms have standard data or file structures and 69 percent have naming conventions. But only 23 percent have data-quality or validation procedures, 19 percent have defined authoritative data sources, 15 percent have designated data owners or stewards, and 14 percent have metadata standards. As data gets reused more widely and feeds automated processes, those gaps compound.
Most striking: only 7 percent of firms say they currently have sufficient capabilities across the areas examined. Among the rest, 79 percent expect to close the gap by training existing employees rather than hiring their way out of it.
What this means for how firms sell engineering
Connected data may change not only how firms work, but what they sell. Eighty-two percent of respondents believe traditional project deliverables and milestones will need to evolve. Seventy-five percent believe connected data and automation will allow firms to create more client value with fewer labor hours. And 72 percent believe traditional hourly pricing makes it difficult to capture the value that creates.
Together, those findings put pressure on the standard commercial model in engineering services. If a firm can deliver more with fewer hours, and it bills by the hour, efficiency becomes self-defeating. Eighty-one percent expect data ownership and reuse rights to become increasingly important contractual issues; 71 percent anticipate opportunities for recurring or lifecycle services beyond traditional project delivery.
The study does not predict which models will win. It does suggest the conditions that produced the current ones are changing.
The long view
The single strongest future expectation in the survey concerns what happens after infrastructure enters service. Eighty-five percent of respondents believe data generated during operations and maintenance will increasingly feed back into future planning and design.
Project data has traditionally been created to support a phase, a deliverable, an assignment. Connected data lets that information keep supporting decisions as an asset moves through construction, operation, maintenance, and eventual renewal, decades during which the software platforms, the firms, and the owners will all change repeatedly.
Which is why the study keeps returning to interoperability rather than to any particular platform. The goal is not technological uniformity, which is neither achievable nor desirable. It is data that keeps its meaning when it moves.
Read the full report at here. From Project Data to Infrastructure Intelligence is part of the ACEC Research Institute’s Firm of the Future series, developed with Strategic Partner Bentley Systems.