BI-Tech: An IoT-Based Behavioral Intervention System for User-Driven Energy Optimization in Commercial Spaces

This study presents the Behavioral Insights and Technology (BI-Tech) IoT platform for closing the intention–action gap via real-time behavioral verification by integrating: 1) synchronized validation of user-reported actions against energy consumption traces, 2) a dual-track evaluation linking macro-scale metered savings with micro-scale 15-minute action windows, and 3) enhancement of Theory of Planned Behavior (TPB) predictive validity under a low-cost deployment (US<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\$}$ </tex-math></inline-formula>75 per node) with off-the-shelf sensors. In a field deployment across 12 small-to-medium commercial offices (SMOs) in Fukuoka, Japan (September–November 2024; N = 57), the platform was evaluated using a dual-track design. Statistical analysis of hour-level observations across valid workdays (baseline: <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$n=511$ </tex-math></inline-formula>; intervention: <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$n=421$ </tex-math></inline-formula>) demonstrated substantial energy reduction with large effect size (Cohen’s <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$d=0.72$ </tex-math></inline-formula>, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$t(930)=10.94$ </tex-math></inline-formula>, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$p\lt 0.001$ </tex-math></inline-formula>), representing a 23.3% decrease in non-HVAC consumption. Targeted evening analysis confirmed intervention effectiveness, with lighting consumption declining 21.4% during notification periods (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$t=2.98$ </tex-math></inline-formula>, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$p\lt 0.01$ </tex-math></inline-formula>, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$d=0.51$ </tex-math></inline-formula>). The action-window analysis isolated user-behavior effects from automated prompts; peak savings of 70.9 Wh were observed within 15 minutes following air-conditioning alerts. The awareness–behavior correlation increased from <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$r=0.336$ </tex-math></inline-formula> (n.s.) to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$r=0.634$ </tex-math></inline-formula> (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$p=0.020$ </tex-math></inline-formula>), <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\Delta r=+0.298$ </tex-math></inline-formula>. In addition, 69.2% of participants shifted toward more conservation-oriented profiles in a four-quadrant behavioral classification. These results demonstrate a validated, low-cost method for real-time verification of IoT-mediated behavioral interventions in offices, emphasizing human-centered strategies rather than infrastructure-intensive automation.