Skip to main content
Research and analysis

Online choice architecture mapping: executive summary

Published 19 November 2026

Background and methodology

Online choice architecture (OCA) refers to how online environments are designed to shape user decision-making. While OCA can support navigation and decision-making, it may also influence behaviour in ways that are not aligned with users’ interests. 

This research aimed to map the prevalence and nature of potentially harmful OCA across a range of e-commerce platforms. The focus was on identifying where and how these features appear within typical shopping journeys, rather than measuring harm directly. 

A total of 630 standardised shopping journeys were replicated across 7 e-commerce platforms, with a spread across different devices, products, and user personas.

Journeys were systematically coded using a structured and iteratively developed codebook of OCA types and sub-types, capturing the presence and characteristics of OCA at each stage of the shopping journey. A proportion of journeys were double-coded to ensure inter-coder consistency.

Findings should be interpreted given certain methodological considerations and limitations. This includes the fact that the standardised shopping journeys were restricted to a predefined set of products, platforms, and journey stages.

There was also some variation arising from platform design and technical constraints. The research also focused on mapping the prevalence and presentation of OCA, rather than measuring user experiences, behavioural impacts, or harm.

Findings

OCA was highly prevalent across all platforms, with an average of 15.7 different types observed per journey. Several OCA types appeared on almost all journeys, including:

  • reviews and ratings (encountered on 100% of journeys)
  • framing (100%)
  • popularity claims (99%)
  • sludge (99%)
  • similar products (99%)

Other OCA varied significantly in its prevalence on different platforms, such as:

  • gamification (ranging from 0% to 99% in prevalence on different platforms)
  • pop-ups and prompts (0% to 87%)
  • partitioned pricing (1% to 84%)

OCA also varied in its prevalence across different stages of the journey. For instance, while reviews and ratings were most prevalent at the ‘product page’ (with 96% prevalence at this stage), feedback was the most prevalent when adding the item to cart (81% prevalence), and at the home page after checkout, personalisation was the OCA encountered most often (49% prevalence).

For some OCA, its prevalence varied significantly on journeys carried out on different devices. For instance, ranking was more prevalent on web journeys (encountered on 100% of web journeys vs. 91% of app journeys), while commitment was more prevalent on app journeys (91% vs. 71% on web).

Although harm was not directly measured, some OCA was perceived by researchers carrying out the coding as potentially more harmful, particularly where they felt overwhelming, misleading, or reduced users’ sense of control. Others appeared to be used in ways that could be primarily beneficial to users of the platform, rather than acting in the best interest of the retailer.

Implications

This research provides a foundation for further work exploring how OCA affects user outcomes. Future research should focus on user experiences and vulnerability factors, to better understand the relationship between OCA practices and potential harm.