Evaluating the relative opportunities of human-based and automated, technology-enabled behaviour change programs

A study by 6t for the ADEME

The ADEME publishes a new study on behaviour change, conducted by 6t

The climate change emergency calls for prompt and large-scale measures. While the necessity to reduce polluting practices such as car-use, overconsumption of disposable plastic items, or high electricity consumption now appear unchallenged, unsustainable practices display a strong inertia (Draetta, 2003; Philipps-Bertin, 2004 ; Rocci, 2007 ; Howarth, 2009). Several barometers published by the ADEME have also shed light on this discrepancy between opinions and daily practices.

Not only does the material context strongly determine social practices by enabling or preventing alternative practices, but many barriers to change also exist at the individual level. In this context, how can one induce the evolution of individual habits, perceptions and skills? How can environmental values be made to transfer into actions? This is the goal of voluntary behaviour change programs.

Voluntary behaviour change programs aim at inducing a sample of beneficiaries to reflect on their own practices, understand their negative impact, and progressively work towards forming a behaviour change goal, and in turn, and effective change of habits. To achieve this goal, these programs use targeted information and personalised advice on a sample of beneficiaries who are in a position to change their practices, that is, live in a context were alternative ways are available and achievable.

These programs have been highly influenced by a wide variety of theoretical models (Theory of Planned Behaviour d’Ajzen, 1991 ; Theory of Interpersonal Behaviour de Triandis, 1977 ; Norm-activation model de Schwartz, 1977 ; Prochaska et Di Clemente, 1986), and experimented in different disciplines, from waste recycling (e.g Viscusi et al., 2011.) to nutrition (e.g. ADEME, 2016 ; Bird et al., 2013), energy sobriety (e.g. Asensio et Delmas, 2016 ; Standdon et al., 2016) to mobility (Rocci, 2007, 2008, 2009 ; Lagadic et al. ; 2019). The latter domain has given rise to largest-scale programs internationally, such as those using the IndiMark® (Social Data/Sustrans) and Travel Blending® (Steer) methods.

Empirical experimentations have proved to be very efficient, but are limited in scale by their strong cost-intensity. A review conducted by Lagadic et al. (2019) and based on programs experimented in France in the field of mobility evaluated the average cost/efficiency ratio at about 973€ per beneficiary, and 2840€ per beneficiary who affectively changed behaviour. These estimations were based on small-scale programs that relied intensively on counsellors-beneficiary interactions to induce change.

How can these progams reach a wider scale? Digital technologies offer new opportunities to automate some functions so as to limit workforce related costs. Research in the field of persuasive technology, which has developed since the 2000s, explore perspectives for replacing human counsellors by entirely digital yet highly personalised tools. What are the opportunities offered by new technologies to generalise these programs and achieve a large-scale impact?

This report presents the result of a three-phase study: first, a literature review covering both theoretical models and empirical evidence related to behaviour change ; second, an in-depth analysis of 5 programs, then compared with an additional 15 cases ; third, operational guidelines to integrate automated, numerical features into behaviour change programs, as well as to better evaluate their efficiency. The report concludes by identifying key avenues for future research.

Key results include the following. Automating behaviour change programs through the use of digital solutions offers a certain number of opportunities :

  • Using automated technological solutions actually allows to present beneficiaires with more personalised contents : the use of complex algorithms allows for the adaptation of targetd information and advice to participants’ practices, constraints, or even their personality. While traditional programs rely on an interpersonal interaction, this interaction is not entirely free, but rather organised around standardised contents the counsellor bases his or her advice on. Human counsellors do not have the same capacity to cross-analyse informations and are thus not able to provide the same level of tayloring that a computer would.
  • Most ambitious VTBC programs integrate a variety of incitative tools. Using new technologies makes the combination of multiple tools easier, thus making the program attractive to a wider sample of users.
  • Behaviour change is structured into different phases, that have been identified in social psychological research. Each phase comes with specific needs ; using technological tools allows to better monitor participants’ progression from one phase to the other. It thus becomes possible to adapt contents to the specific stage in which the individual user finds himself/herself at a given time, thus optimising efficiency.
  • Using digital technologies offers prospects for developing tools that have been little explored in traditional, human-based programs. This is especially the case for gamification, as well as real-time peer-to-peer comparison.
  • Using digital technologies eases data collection and processing. This not only optimises the use of traditional tools such as feedback, but can also greatly improve program evaluation.
  • Once the digital tool is created, its diffusion to an extra participant (marginal cost) is close to none (to the exception of data storage and API fees). Technology thus makes large-scale deployment possible at a reasonable cost.

A certain number of risks should also be kept in mind :

  • The use of technological tools may evict certain publics, especially the elder and as well as precarious households. On-demand human counselling may be a solution to ensure the inclusivity of these programs.
  • While little data is available on the relative efficiency of human-based and technology-based recruitment, interviews conducted with key actors in the sector suggest that door-to-door recruitment, as well as the organisation of real-life events, may ease recruitment and ensure a variety of profiles.
  • Key governance challenges have been identified : many programs did not come to fruition due to a difficult coordination between local governments, private firms and academic research teams. Similarly, the great heterogeneity of evaluation protocols, and the lack of a rigorous evaluation in may cases, limits the progress of reseach in the sector.

This report thus reach a certain number of operational recommandations, including :

  • Privileging deployment at a scale that allows users to be acquainted with each other, so as to create a social dynamic around the program. Large-scale deployment is achieved through a multiplication of user communities ;
  • Exploring the opportunities offered in the sector of mobility – travel planners used to optimise the issuance of personalised advice – and energy – use of automatically collected data for real-time feedback)
  • Using technology at the pre-action stage (the beneficiary is willing to change but has not yet identified a suitable solution) to provide the most personalised advice, as well as in the post-action stage (the beneficiary has changed his behaviour) to ensure maintenance in the long-run. This last-phase is little tackled in traditional, human-based programs.
  • Developing those tools that were little efficient in human-based programs : gamification and real-time peer-to-peer comparison
  • Seizing the opportunity offered by digital technologies to improve and standardise evaluation methods, so as to move VTBC research forward.

The full report and its synthesis (in French) can be found on the ADEME website :
Share this publication :