Changing behaviours through digital health communications

Published on 18 September 2024 | Simon Moore

Evaluating the contribution digital communications interventions make to improvements in behavioural determinants and changes to behavioural intention remains challenging. Novel approaches are being used, such as the acyclical behaviour change diagram (ABCD) method, but there are few examples of how this approach has been applied to living brands.

This 12-month project applied the ABCD approach to Avert’s Be in the KNOW sexual health information brand, which reached over 13m people across Africa in 2023. The project aimed to understand and improve health learning journeys on Be in the KNOW; develop evidence on whether the brand impacts knowledge, self-efficacy, attitudes and behavioural intention; and assess the value of the ABCD approach to underpin digital evaluation at a programme level.

The project began with a theoretical analysis of the behavioural aims of Be in the KNOW website and social media content using the ABCD method (summarised in our previous blog). This involved selecting representative web pages and social media posts, breaking down these ‘interventions’ into causal-structural chains – mapping from behaviour change methods used, through change objectives, and behavioural determinants (such as knowledge, attitudes, and risk perception), to overall behavioural aims – and visualizing these intended aims through ABCD diagrams. The Intervention Mapping Taxonomy was used to guide the process, ensuring a systematic evaluation of the behavioural components of the content.

The analysis revealed the change objectives, determinants, and behavioural aims for each webpage and post, as well as identifying the behaviour change methods used, supporting assumptions about their effectiveness, and exploring opportunities for improvement. By comparing the results of this analysis to actual user behaviour (from analytics, surveys and UX sessions), improvements were implemented. Four months of post-improvement data was collected, alongside endline surveys and interviews to evaluate the impact of interaction with Be in the KNOW on behavioural determinants and behavioural intention.

Endline data provided strong indications that Be in the KNOW influences behavioural determinants, supports specific change objectives and impacts behavioural intention.

Results

Endline analytics suggested that this behaviour change approach using ABCD analysis:

(1) guided evidence-based intervention improvements. The ABCD process highlighted in detail the key change objectives, determinants and sub-behaviours for each webpage and post. The process highlighted things that are missing, areas where we could make improvements, and what we needed to measure in order to see and explain if expected behaviour has happened. It also helped institutionalise some of the behavioural analysis of this process’ into ongoing new content development and content review processes.

(2) increased user engagement (a proxy for behavioural determinants such as knowledge). Overall engagement levels were higher at endline with higher percentages of people finding the two pages useful and higher page engagement rates. At endline a higher percentage of users were clicking one of the options for further relevant exploration. A high percentage of survey participants said it is easy to find what you are looking for (86.2%), that content is relevant (99.6%), and that it is presented in the right level of detail (68.1%). On social media, overall engagement had increased slightly over the endline period on both Facebook and Instagram, but the changes were more marginal.

(3) improved users’ learning journeys. Changes made as a result of the ABCD analysis, including new quiz content, changes to presentation of personal stories, additional internal links, and more visibility for discussion starter content, fed through into improvements in user journeys at endline (increased traffic to quizzes, personal stories and discussion starters). However, there was no increase in onward journeys to modern contraception content.

Endline self-report survey and user interview data provided strong indications that Be in the KNOW:

(4) influences behavioural determinants such as knowledge, self-efficacy, and attitudes.

(5) supports specific change objectives identified through the ABCD analysis. For example, high percentages of respondents said interacting with Be in the KNOW had made them believe more strongly in the importance of using modern contraception (94%), getting tested for HIV if you’ve been exposed (96.4%), expecting to have sex that is pleasurable and safe (90.5%), and not feeling any shame about testing for HIV (95.7%).

(6) impacts behavioural intention (such as getting tested or talking to a partner). Overall, 94.8% said they were more likely to look after their sexual health (vaginal sex page) and 92.8% more likely to get tested for HIV (when to test page), with similarly high levels of intention on more specific behavioural aims of the pages identified through the ABCD analysis.

A useful theory-driven approach

While live online brands cannot be a controlled environment, the project demonstrated that ABCD analysis provides a useful, theory-driven approach to underpin website evaluation and improvement. The approach is an iterative process not a one-off activity, with each stage of the process providing new insights to be tested in future iterations, so taking this approach needs allocation of ongoing time and resources. Working with a live product also creates a number of data challenges, including the impact of external factors such as Google algorithm changes, and variability factors created by other changes associated with a live brand.

Despite these broader challenges around digital evaluation, the project provided evidence that taking a behaviour change approach using ABCD analysis can helpfully guide iterative changes and increase user engagement and improve users’ learning journeys.

Simon Moore is Director of Programmes and Digital Health Strategy at Avert. A summary of the research can be downloaded here.

Thanks to Gido Metz from Maastricht University for his input and advice throughout the project and to Gilead Sciences for their funding of this work.