The
Confidence Gap

Commercial real estate leaders are not short on confidence. Eighty-four percent say they are confident in the accuracy of their CapEx budgets when those budgets are approved. Ninety-two percent are confident in their organization's ability to make capital investment decisions using accurate, timely information.

The outcomes are less predictable. Sixty-six percent say completed CapEx projects typically exceed budget by more than 5%. Thirty-nine percent typically exceed budget by more than 10%. And 91% say incomplete, delayed, or siloed CapEx data has caused their organization to revise or reverse a capital allocation decision in the past 24 months.

These findings aren't a reflection of a lack of expertise. CRE teams have spent years building the processes, technology, and institutional knowledge needed to manage complex capital programs. They know how to pressure-test budgets, refine scope, and adjust when conditions change.

They are also very good at compensating when the infrastructure connecting those functions does not. Teams reconcile spreadsheets, track down missing property information, re-enter approved budgets, call vendors for context, rebuild forecasts when scope moves, and use institutional knowledge to fill the spaces between systems.

We call the distance between those two things the Confidence Gap: the distance between how confident CRE leaders feel when making capital decisions and how much control their operating infrastructure gives them over what happens next.

The gap matters because capital decisions do not happen once. They travel. A diligence assumption becomes a capital plan. A plan becomes a budget. A budget becomes a scope. A scope becomes a bid. A bid becomes a project. A project becomes an actual cost and, ideally, a measurable financial outcome. At every transition, context can be preserved, or it can be lost.

Seen separately, these findings look like a property-data problem, a budgeting problem, a project-management problem, a reporting problem, and an AI-readiness problem. Taken together, they point to a more fundamental issue: information, decisions, people, and workflows do not consistently maintain continuity as capital moves through the lifecycle.

Before an organization can orchestrate capital, it has to know what it is orchestrating. Today, teams assemble capital plans from a broad mix of systems and sources.

Using multiple sources is not a problem. The question is whether they describe the asset consistently enough to support a reliable capital decision.

These findings change where the industry should look when trying to understand capital performance. When a project exceeds budget, attention naturally turns to execution: contractor pricing, change orders, schedules, or material costs. Some execution problems, however, may be upstream information problems arriving late.

There is no single bad-data culprit. The percentages are close, and that is exactly the insight. The problem is distributed across the underlying property record. What is in the building? What condition is it in? How old is it? How much of it is there? What has already been replaced? What did similar work cost? What has already been approved or spent? Different answers may live in different places or with different people.

Portfolio scale turns small information gaps into operating exposure. One missing equipment record is an inconvenience. An inconsistent asset record repeated across hundreds or thousands of units can distort quantities, replacement assumptions, scope, and timing.

Scope surprises often begin as context-transfer problems. Among respondents who cite scope surprises as a source of variance, the leading causes include stakeholder changes after work begins and contractor or vendor interpretation diverging from the intended scope, both at about 24%. Hidden conditions follow at 22%, with incomplete scope definition before work begins at about 20%.

Several of these are not simply construction problems. They are context-transfer problems: what ownership intended, what asset management approved, what construction scoped, what the contractor understood, and what existed at the property.

Execution is where the effects of upstream uncertainty and broken handoffs become easiest to see. Scope drift is one of the clearest signals. Seventy-six percent experience meaningful scope drift on at least 10% of projects. Fifty-eight percent say scope drift occurs on 10 to 29% of projects. About 14% experience it on 30 to 59% of projects, and another 3% report it on 60% or more. Across the sample, the average is roughly one in five projects.

Scope changes can be legitimate. Buildings reveal conditions that were not visible in diligence. Ownership priorities shift. Code requirements change. The question is how often a change represents new information versus information that existed but did not travel with the project.

The survey's causes of scope surprise suggest both are happening. Hidden conditions matter, but so do stakeholder changes after work begins, divergent vendor interpretation, and incomplete scope definition.

Why the gap matters economically. The Confidence Gap is not only an operational inconvenience. It changes the quality and speed of capital allocation. Closing it creates several forms of economic upside:

The research does not quantify the dollar value of each benefit, so it would be wrong to turn these findings into a universal ROI claim. What it does show is how frequently the conditions that create rework, reversals, and variance appear across the market.

The opportunity is to move intervention earlier. If a team can identify missing property information before a budget is approved, preserve assumptions through scope and bidding, and see variance as it emerges, it has more options than a team discovering the same issue after capital has already moved.

When asked what would make them comfortable allowing AI to autonomously make or execute CapEx decisions, respondents pointed to safeguards across the decision-making process:

The message isn't that CRE is resistant to autonomous technology. It's that autonomy has to be earned.

The organizations best positioned for the next wave of AI will be those that can give intelligent systems reliable information about the asset, connect that intelligence to capital decisions, preserve context as projects move into execution, and maintain visibility into what changed and why.

That is a fundamentally different challenge from simply adding AI to another point solution.

The research does not suggest an industry incapable of managing capital. Quite the opposite. CRE has built sophisticated organizations staffed by experienced people who have learned how to make complex capital decisions despite fragmented information, disconnected systems, and imperfect handoffs.

The cost of that compensation is becoming more visible. It appears in budgets that move after approval, decisions that must be revisited, duplicated tracking, scope drift, late variance detection, and completed-project knowledge that is difficult to reuse.

AI raises both urgency and opportunity. Boards and investors want adoption. Capital teams already see concrete use cases in reporting, prioritization, predictive replacement planning, budgeting, and variance detection. But more autonomous workflows will require a stronger foundation of verified data, connected context, auditability, and governance.

The organizations that lead the next era will not necessarily be the ones with the most software, the largest datasets, or the greatest number of AI tools. They will be the ones that can preserve intelligence and intent as capital moves from property to plan, from plan to execution, and from execution back into learning.

The research was conducted by Censuswide between July 23 and August 6, 2026, among 306 U.S.-based commercial real estate decisionmakers and senior operators age 30 and older. Respondents were directly involved in functions including Investments/Acquisitions, Asset Management, Capital Projects/Construction Management, Property Operations, and Finance/Accounting.

The sample represents organizations with meaningful capital programs. Approximately 54% report annual CapEx of $100 million to $499.99 million, and about 9% report $500 million or more. The mean annual CapEx represented in the sample is approximately $226 million.

Some questions allowed multiple responses, so percentages may total more than 100%. Asset-class cross-tabs can overlap because respondents may operate across more than one asset class. Percentages may not sum to exactly 100% because of rounding. Small-base cuts should be interpreted directionally.

Censuswide is a member of the Market Research Society and the British Polling Council and is a signatory to the Global Data Quality Pledge.