Shopping may appear to be a simple transaction, but every purchase is connected to a complex chain of freight transport, packaging, warehousing, delivery, store operations, digital activity, and consumer travel. A new study published in Artificial Intelligence & Environment introduces a standardized method for measuring the greenhouse gas and air pollutant emissions generated across this entire retail journey.
The protocol covers both online and traditional in-store shopping and provides nationwide parameters for China from 1990 to 2023 across 16 major merchandise categories. It is designed to make environmental assessments more comprehensive, transparent, and comparable across products, regions, shopping methods, and time periods.
"Shopping-related emissions are often evaluated using only a few visible activities, while important sources such as air freight, packaging materials, product returns, and consumer travel may be overlooked," said corresponding author Ruibin Xu. "Our protocol provides a consistent way to follow emissions through the full retail supply chain and identify where meaningful reductions can be achieved."
Following emissions through the full shopping journey
The researchers divided shopping into six major supply chain phases: wholesaler transport, long-distance transport, last-mile delivery or consumer shopping trips, packaging use, store or electronic device operations, and storage.
The framework evaluates greenhouse gases, including carbon dioxide, methane, and nitrous oxide, as well as major air pollutants such as fine particulate matter, nitrogen oxides, sulfur dioxide, carbon monoxide, ammonia, black carbon, organic carbon, and non-methane volatile organic compounds.
Unlike approaches that consider only trucks, cars, or cardboard boxes, the new protocol incorporates multiple freight and travel modes, including airplanes, trains, ships, vans, private cars, public transport, motorcycles, electric bicycles, walking, and cycling. It also accounts for eight types of packaging used in online retail and uses Monte Carlo simulations to estimate uncertainty.
Online and in-store shopping have different emission hotspots
The researchers demonstrated the protocol using shopping for necessities, textiles, and foodstuffs in China. In 2023, total greenhouse gas emissions associated with these categories were estimated at 11 million metric tons for necessities, 8.2 million metric tons for textiles, and 23 million metric tons for foodstuffs.
Online shopping generated the larger share of emissions for necessities and textiles, contributing approximately 60% and 77%, respectively. Traditional shopping produced the larger share for foodstuffs, accounting for about 74%.
For online purchases, long-distance freight and packaging were the dominant greenhouse gas sources. Packaging alone contributed 54% of online shopping emissions for necessities and 70% for textiles. For traditional shopping, consumer trips to stores dominated emissions, accounting for 80% for necessities, 77% for textiles, and 70% for foodstuffs.
The analysis also showed why simplified assessments can substantially underestimate retail emissions. When air freight, diverse packaging materials, and multiple transport modes were included, estimated greenhouse gas emissions were approximately 50% to 400% higher than results produced using a more limited earlier method.
Three decades of changing retail emissions
By incorporating historical changes in freight systems, vehicle use, delivery efficiency, electricity, and emission standards, the protocol can track how shopping-related pollution evolves over time.
From 1990 to 2023, greenhouse gas emissions increased sharply across all three case-study categories as consumer demand expanded. However, stricter vehicle emission standards helped produce temporary declines in several air pollutants during certain periods.
These findings suggest that cleaner freight vehicles, lower-emission transport, reduced air freight, efficient delivery networks, lighter and more sustainable packaging, and fewer car-dependent shopping trips could all support greener retail systems.
Providing reliable environmental logic for AI
The team also developed an agent-based model that applies the protocol at the level of individual purchasing events. The model can simulate different consumer behaviors and supply chain configurations while tracing emissions through all six phases.
By providing structured data and verifiable calculation logic, the framework could serve as a reliable reference for future artificial intelligence systems that evaluate retail emissions and test sustainable shopping scenarios.
The researchers hope the protocol will support scientists, businesses, policymakers, and consumers in moving beyond simple comparisons between online and in-store shopping and toward targeted, evidence-based strategies for reducing the environmental footprint of retail.
===
Journal reference: Xu R; Shen H; Wang P; et al. A protocol for the comprehensive evaluation of greenhouse gas and air pollutant emissions from shopping. AI Environ. 2026, 1(2): 106-119. DOI: 10.66178/aie-0026-0013
https://www.the-newpress.com/aie/article/doi/10.66178/aie-0026-0013
===
About the Journal:
Artificial Intelligence & Environment is an international multidisciplinary platform for communicating advances in fundamental and applied research on the intersection of environmental science and artificial intelligence (AI). It is dedicated to serving as an innovative, efficient and professional platform for researchers in the cross-discipline fields of earth and environmental sciences, big data science and AI around the world to deliver findings from this rapidly expanding field of science. It is a peer-reviewed, open-access journal that publishes critical review, original research, rapid communication, view-point, commentary and perspective papers.
