.png)
66% of CRE professionals use AI daily or weekly. Only 5% trust it enough to inform an actual real estate decision (First American Data and Analytics, via ICSC, June 2026).
That gap is the story of where the industry is right now. Not whether AI is useful. Not whether it's being adopted. Whether anyone has figured out how to make it produce something that changes how a decision gets made.
The answer, for most properties, is not yet. But the reasons why are specific and fixable.
The Implementation Gap
88% of CRE property owners and investors were running AI pilots in 2025. Only 5% had achieved all their program goals (JLL Global Real Estate Technology Survey, 1,500-plus respondents, October 2025). 42% of companies scrapped most of their AI initiatives in 2025, more than double the 17% that did so in 2024. The average organization abandoned 46% of its AI proofs-of-concept (S&P Global Market Intelligence, May 2025).
These aren't small organizations with limited resources. 96% of institutional CRE investors plan to increase AI spending in the next year. 87% have already grown their technology budgets because of AI (JLL, 2025). The problem isn't investment. It's sequencing.
The Real Reason Most Initiatives Stall
RAND Corporation published one of the most rigorous analyses of AI project failure available: 65 semi-structured interviews with experienced AI practitioners across industries. The finding that matters most for retail property teams is this: 60% of practitioners cited data quality as a root cause of failure. Leadership failures, specifically defining the wrong problem and setting the wrong metrics, were cited by 84%.
RAND's verbatim: "80% of AI is the dirty work of data engineering. You need good people doing the dirty work; otherwise their mistakes poison the algorithms" (RAND Corporation, August 2024).
Gartner put a number on the consequence: organizations will abandon 60% of AI projects that lack AI-ready data through 2026 (Gartner, February 2025).
Here's the Canadian-specific version of that problem. PwC and ULI Canada's 2026 Emerging Trends report interviewed Canadian CRE professionals and found a recurring theme: "lack of clean, integrated data... fragmented and outdated property management systems limit the effectiveness of AI for analytics and forecasting." Quebec organizations that invested in centralized data governance frameworks ahead of Law 25 compliance ended up with a structural AI readiness advantage. That outcome wasn't accidental.
And here's the part that should bother every property team that thinks they're ahead of this: Dealpath's 2026 survey found that 83% of institutional CRE firms rate their own data as "AI-ready." 90% of those same firms say data is limiting AI's impact. That self-assessment gap is what the industry is actually navigating right now.
What's Actually Working
The implementations producing measurable outcomes share one characteristic: they started with a clearly defined operational problem and built the data infrastructure to solve it before selecting the technology.
Cadillac Fairview is the clearest Canadian example. In October 2023, CF deployed an AI-powered energy intelligence platform across 9.5 million square feet of office space, integrating into existing Building Automation Systems to autonomously adjust HVAC settings using weather data, occupancy patterns, and heating and cooling loads. By early 2026, the results were documented: 4 to 6% annual energy savings on an energy budget exceeding $110 million, and a 10 to 15% reduction in maintenance work orders.
Software licensing cost: approximately $0.02 per square foot per year. CF's SVP of Operations described the change management across property teams, data and technology, and operations services as "the most important part of the project" (Globe and Mail, February 2026).
Brixmor Property Group in the U.S. started differently. Before deploying any external AI tools, they built Chat BRX, an in-house generative AI tool built specifically to expose employees to AI on low-risk, immediately useful tasks: lease clause summaries and marketing materials. Once staff were comfortable, they layered in Copilot and Claude. The result over two years: lease negotiation timelines trimmed by approximately 15% (CEO Brian Finnegan, NYU Annual REIT Symposium, Spring 2025, via ICSC, June 2026).
Kimco Realty used AI to process approximately 200,000 pages of merger documents, saving roughly three weeks of manual work. Simple application. Immediate value. No complex integration required.
The pattern across every working implementation: start with a specific problem, not with the technology.
The Three-Phase Approach
ICSC's implementation roadmap for shopping centres (December 2025) and BCG's analysis of 14 leading mall operators (October 2025) point to the same sequencing.
Phase one is foundational: generative AI applied to content production, document processing, and internal communications. No data integration required. Immediate productivity gains. This is where staff get comfortable with AI before the organization asks AI to do anything consequential. Harvard Business School and BCG research found that employees using AI completed 12% more tasks, finished them 25% faster, and produced work rated 40% higher in quality, with the largest gains going to lower-performing staff (cited in NAIOP Development Magazine, Summer 2026).
Phase two is marketing and customer-facing: personalized email and customer segmentation, AI-powered chatbots for shopper queries, and digital marketing automation. These require consolidated CRM and campaign data, which is why the data unification work in phase one is a prerequisite, not a parallel track. BCG documented a 30% improvement in conversion rates across mall operators using automated segmentation and targeted digital marketing.
Phase three is operational: AI-powered foot traffic and dwell-time analytics, predictive maintenance, energy management. ICSC and Purdue University found malls using AI-driven insights seeing 11% traffic improvements. BCG documented 30% average revenue increases for pilot tenants at malls using AI-powered footfall tracking.
ICSC's assessment of the timing: "Shopping centres implementing AI strategies today will have 12 to 18 months of data, optimization experience, and established tenant relationships by the time the majority of the industry recognises this as a competitive necessity" (ICSC, December 2025).
The hype phase is over. The properties that spent the last two years in meetings about AI are now 12 to 18 months behind the ones that started building.
Interested in what an AI implementation roadmap looks like for your property?
