Two social features, opposite results: a diet app trial argues against "add engagement"

Researchers at Carnegie Mellon, Hofstra and UC Irvine, working with a Pittsburgh-based nutrition-focused mobile health company, followed nearly 350 adults for four months in 2014 and found that two socially delivered persuasive features moved behaviour in opposite directions ^[3][4]. The work is published in the International Journal of Information Management ^[2], and its practical content is a warning to any team that treats "social" or "engagement mechanics" as a single interchangeable category.

The two features were narrow and comparable. One was photo-informed expert feedback: proactive, personalised comment from a dietitian on photos of a user's meals ^[5]. The other was photo-based peer observation: users seeing other people's shared meal posts on a photo-based social platform ^[6].

Expert feedback increased app use, slowed disengagement over time, and improved meal recording, but did not influence meal balance ^[7]. Peer observation went the other way: it reduced engagement, primarily among relatively unhealthy eaters, and did not significantly affect food choices ^[8]. Same product, same medium, same photographs, opposite sign on the engagement metric ^[11]. The authors' reading of the peer result is that some people may be deterred by the relatively healthy norms of a photo-sharing environment ^[9], which is the uncomfortable case for a health product: the feature discourages the segment with the most to gain.

Note also what neither feature did. Expert feedback did not shift meal balance and peer observation did not shift food choices, so across both arms the measured dietary outcome stayed put ^[12]. The honest summary is that one feature bought self-monitoring behaviour that the researchers suggest may sustain general consumers who value authoritative guidance, without an immediate translation into dietary balance ^[7]. Retention is the industry's standing problem here: diet-related apps make up about 12% of the health app market and struggle to keep users ^[1][10].

Yi-Chin Kato-Lin of Hofstra, who led the study, draws the distinction that matters for design: strategies such as gamification, goal setting and self-monitoring are focused on individuals, while peer- and expert-based persuasion involve interpersonal social interactions whose dynamics can foster a kind of social contagion of health behaviours ^[13]. Rema Padman of Carnegie Mellon's Heinz College, a co-author, says that although diet apps reach a wide range of people, "we know little about their effects on users' eating behaviors" ^[14]. The theoretical scaffolding was social norms theory, media richness theory and social comparison theory ^[4].

The limits are stated plainly by the authors. The two features were not perfectly symmetrical, so the results may not generalise to other expert or peer designs; participants were mostly young Android users in the United States already interested in healthy eating; and the data are from 2014, since when digital health technology has changed considerably ^[15]. Julie Downs of Carnegie Mellon argues the features "reflect underlying patterns of human behavior that persist despite rapid technological changes" ^[16], which is a claim about durability rather than evidence of it.

What to watch: whether anyone replicates the peer-observation result on a current cohort and at scale, and whether app teams start reporting engagement effects split by baseline behaviour rather than in aggregate. A feature that lifts your average session count while shedding your least healthy users looks fine on a dashboard ^[8][11].