Background Online grocery shopping is becoming increasingly important; however, it remains unclear which interventions are effective in promoting not only healthier food choices but also more sustainable food choices within this environment.

We set out to 1) understand how we can use a browser extension to collect data on online food purchases and implement and evaluate interventions, and 2) collect preliminary data on recruitment, data collection, and the randomisation process to inform a larger trial investigating the effectiveness of eco-labels and other interventions promoting sustainable food purchases within an online grocery environment in the UK.

Methods We conducted a pilot parallel 2-arm randomised controlled trial with a 1:1 participant allocation ratio in a real-life online UK supermarket from September to October 2023. A browser extension was used to implement the intervention on the supermarket website and collect participants’ grocery data. Participants recruited from an online research agency were randomised to complete their normal grocery shopping under one of two conditions: eco-labels providing participants with information on the environmental impact of their food purchases, scored from A (best) to G (worst) vs. no eco-labels. The trial had a 2-week baseline data collection phase during which the eco-labels were off for both groups, followed by a 2-week intervention period during which the eco-labels were turned on for only the intervention group. The primary outcome of our full trial will be the average eco-score of baskets. Differences between the control and intervention groups were checked using t-tests. Participants also completed surveys at baseline and endline reporting their characteristics and experience of participating using the browser extension.

Results A total of 504 participants were screened, and 161 met the eligibility criteria and were invited to participate in the study. Out of the 161 invited participants, 57 downloaded the browser extension. The sample size in the pilot was too small to find significant differences between the control and intervention groups in the eco-score of shopping baskets [32 ± 13 vs. 41 ± 14 (mean difference, MD, -9.5; p = 0.22)]. To improve recruitment rates in the main trial, responses from 61 participants in the endline survey suggested technical support, reminders, greater incentives and more information about the eco-labels were needed.

Conclusion Our study revealed that participants required more support, reminders and incentives to download the browser extension, and shop using it. These lessons learned will improve the design and implementation of the larger-scale trial.