Does packaging labelling change behaviour?

Two different claims

Labels are used on the assumption that printed information changes what people do. That assumption splits into two questions. Does a disposal instruction change how people sort? And do environmental marks change what they buy?

The sorting evidence is modest and qualified. The purchasing evidence is weaker than marketing surveys imply. Real-world purchase shifts are single-digit. They sit an order of magnitude below the rates consumers report. The effects are real and small. Studies that have measured both find the same gap.

Each study below is recorded with its design, sample, reported effect and a strength grade.

Sorting

The main controlled test of a disposal instruction is Allison et al. (2024). A UCL team, including the leads of the Big Compost Experiment, ran an online experiment with 1,008 UK participants. They sorted eleven compostable packaging formats into general waste, food waste or recycling under six instruction conditions. With no label, people “disposed of items intuitively based on appearance”: packs that looked recyclable went to recycling regardless of what they were. A “do not recycle” label “increased correct general waste disposal by 11.15 percentage points overall” (OR = 0.09, 95% CI [0.08, 0.10], p < 0.0001). The effect held for most formats, not all. For routing compostables to food waste, “compost with food waste” and “put with food waste” outperformed the bare word “compost” (Allison et al., 2024). Strength grade: moderate. Conditions were randomised and the sample is large for an online task, but the outcome is simulated sorting, not observed disposal. The sample also over-represents composting-engaged people. The result describes attentive consumers under instruction, not unprompted household disposal.

The study that measured disposal with actual weights did not test an on-pack label. Linder, Lindahl and Borgström (2018) ran a two-year field experiment in Stockholm multi-family housing: 474 households, waste weighed at nine stations. The intervention was a leaflet about food-waste sorting. Food waste recycled rose by about 26 per cent against the pre-intervention average, and the difference was still there eight months later (Linder et al., 2018). Strength grade: strong for observed behaviour, objective weights and a control group — but for a leaflet, not a pack mark. It shows that sorting information can change household behaviour. No on-pack study meeting the same measurement standard was found.

Industry evidence sits between those two. WRAP’s on-pack trial put a front-of-pack “most people recycle me” message, with the Recycle Now logo, on Boots and Radox bathroom bottles. Four groups, 4,000 participants. Of those who saw the label, 91 per cent said they put the bottle in recycling, against 86 per cent of those who did not (WRAP, 2020; 2021). Strength grade: weak-to-moderate. The sample is large, but the outcome is self-reported. There was no control site and no objective measure.

No peer-reviewed field experiment of an on-pack disposal label with objectively measured sorting was found. The located evidence is simulated sorting, self-report, and one field trial of a different instrument. The sorting claim holds on that evidence. The study that would settle it has not been done.

Purchasing

The purchasing literature is larger. Stated willingness is high. Hypothetical-choice effects are positive. Measured real-purchase effects are positive but small.

Potter et al. (2021), in Environment and Behavior, reviewed 56 studies, 42,768 participants and 76 interventions on environmental labels and food selection. “[S]ixty comparisons favored the intervention and 16 favored control.” The authors conclude that ecolabelling “was associated with the selection and purchase of more sustainable food products” (Potter et al., 2021). The review used vote counting, not a pooled effect size. Of 56 studies, 41 ran in hypothetical settings and 15 in actual ones. It establishes direction, not magnitude. Strength grade: moderate for direction; uninformative for size.

Field experiments supply the magnitudes, and they sit in a narrow range. Elofsson and colleagues used scanner data from 17 Swedish grocery stores. Climate labelling “increased demand for medium-fat climate labelled milk by approximately 7%.” They add: “The response is significantly smaller than suggested by consumer surveys but larger than that observed in earlier studies of actual purchasing behaviour” (Elofsson et al., 2016). Brunner et al. (2018) put traffic-light carbon labels on a Swedish university restaurant. Green-labelled meat dishes rose 11.5 per cent, red-labelled fell 4.8 per cent, and total emissions fell 3.6 per cent. Vlaeminck, Jiang and Vranken (2014), in an incentive-compatible experimental supermarket, found an environmental label raised the eco-friendliness of food choices by about 5.3 per cent. Strength grades: strong (Elofsson — real stores, real sales); moderate (Brunner — one site, no control site); moderate (Vlaeminck — constructed store).

Across these studies, real-setting effects are positive, statistically significant, and single-digit: roughly 4 to 7 per cent relative shifts on targeted products. System-level outcomes, such as Brunner’s 3.6 per cent emissions cut, are smaller still. An earlier draft of this library called the purchasing effect “statistically indistinguishable from zero” in physical retail. The located studies do not support that phrasing. What they do support is the gap between measured and reported behaviour. Flash Eurobarometer 535 reports 60 per cent of EU citizens saying they have bought products specifically for lower environmental impact (European Commission, 2023). That stated rate is an order of magnitude above every measured field effect above.

Why sorting and purchasing differ

At the bin, the pack is in hand. The question is where it goes. The label is often the only information present, and the act costs seconds. Allison et al.’s 11-point effect of an explicit negative instruction fits that setting. At the shelf, the mark competes with price, brand, appearance and habit — the attention contest in the next article. For most buyers the environmental attribute is one consideration among many. Single-digit purchase effects fit that competition.

Regimes that expect labels to sell sustainability rest on the weaker claim. Regimes that use labels to route packaging rest on the better-evidenced one.

What a small effect can still mean

A label acts per pack encounter, and encounters number in the billions. An 11-point improvement in correct disposal, or a 5-point self-reported recycling lift, would move real tonnage if the experimental effects held among inattentive and unengaged consumers. That is the external-validity question the simulated and self-reported designs leave open. Public-health labelling often works the same way: low unit cost, wide coverage, small per-contact effect.

Larger measured behaviour change in adjacent work comes from other instruments. Cart-level feedback cut contamination by 57 per cent in a pilot and 19 per cent citywide, measured by physical audit. Deposit systems move return behaviour by majorities, through price rather than information. Against those comparators, single-digit label effects look like a supporting layer, not the main driver. The comparison is limited by different target behaviours and measurement standards. Where price or personalised feedback has been tested alongside information, information produces the smaller effect.

No cost-effectiveness comparison for a packaging label was found. Rapid reviews are explicit that little is known about effectiveness of contamination-relevant interventions, “and even less about cost effectiveness.” Labelling may be cheap per contact and expensive per tonne diverted. No study estimating either quantity was found.

What the evidence does not cover

The sorting evidence covers instructed, attentive, largely engaged participants — not the distracted, the rushed, or the people excluded from label access in a later article. The purchasing evidence is almost all food and drink, short exposure, and single-label introductions — not a multi-mark shelf, and not durable goods. Neither literature measures persistence at scale (Linder’s eight-month follow-up is the exception). Neither connects its behavioural endpoint to capture rates, contamination or facility economics. That gap is the subject of What is not known?

The grades “modest” and “conditional” describe the located evidence. They are not a dismissal of the instrument. The studies are consistent with labelling as one tool among several: used at the moment of disposal or purchase, worded as an instruction rather than a symbol, and costed against alternatives. No jurisdiction reviewed for these articles states that position in those terms.

Packaging labelling changes behaviour by measured and small amounts, mainly where the decision is taken with the label in view. Larger effects are not established on the current record.

References

Allison, A.L., Băitanu, A., Purkiss, D., Lorencatto, F., Michie, S. and Miodownik, M. (2024) 'Enabling desired disposal of compostable plastic packaging: an evaluation of disposal instruction labels', Frontiers in Sustainability, 5, 1376519. Available at: Open source (Accessed: 18 August 2026).

Brunner, F., Kurz, V., Bryngelsson, D. and Hedenus, F. (2018) 'Carbon label at a university restaurant — label implementation and evaluation', Ecological Economics, 146, pp. 658–667. Abstract available at: Open source (Accessed: 18 August 2026).

Elofsson, K., Bengtsson, N., Matsdotter, E. and Arntyr, J. (2016) 'The impact of climate information on milk demand: evidence from a field experiment', Food Policy, 58, pp. 14–23. Working-paper abstract available at: Open source (Accessed: 18 August 2026).

European Commission (2023) EU Ecolabel use and recognition rise, new Commission data shows (Flash Eurobarometer 535), 12 October. Available at: Open source (Accessed: 18 August 2026).

Linder, N., Lindahl, T. and Borgström, S. (2018) 'Using behavioural insights to promote food waste recycling in urban households — evidence from a longitudinal field experiment', Frontiers in Psychology, 9, 352. Available at: Open source (Accessed: 18 August 2026).

Potter, C., Bastounis, A., Hartmann-Boyce, J., Stewart, C., Frie, K., Tudor, K., Bianchi, F., Cartwright, E., Cook, B., Rayner, M. and Jebb, S.A. (2021) 'The effects of environmental sustainability labels on selection, purchase, and consumption of food and drink products: a systematic review', Environment and Behavior, 53(8). Available at: Open source (Accessed: 18 August 2026).

Vlaeminck, P., Jiang, T. and Vranken, L. (2014) 'Food labeling and eco-friendly consumption: experimental evidence from a Belgian supermarket', Ecological Economics, 108, pp. 180–190. Abstract available at: Open source (Accessed: 18 August 2026).

WRAP (2020) On-pack labelling and citizen recycling behaviour. Report mirror available at: Open source (Accessed: 18 August 2026).

WRAP (2021) One simple message on pack could drive up recycling rates, press release, September. Available at: Open source (Accessed: 18 August 2026).

Note on sources and verification

Allison et al.'s figures — the sample of 1,008, the 11.15-percentage-point effect and its reported statistics, the appearance-driven baseline, and the instruction-wording comparisons — are from the open-access article as retrieved; the odds ratio as printed appears to index reduced odds of mis-disposal and should be read against the full text. Linder et al.'s design, sample and effect (approximately 26 per cent, persistence at eight months) are from the open-access article; it is a leaflet intervention and is expressly not cited here as an on-pack study. WRAP's trial figures (91 versus 86 per cent, n = 4,000, self-reported) are from the report and press release cited. Potter et al.'s counts (56 studies, N = 42,768, 60 versus 16 comparisons, 15 actual versus 41 hypothetical settings) are from the journal page and the review's public summaries; the review computed no pooled effect, and this article accordingly attributes direction only. Brunner et al.'s and Vlaeminck et al.'s findings are quoted from publisher abstracts; full texts were not retrieved. Elofsson et al.'s 7 per cent and the "significantly smaller than suggested by consumer surveys" sentence are from the working-paper abstract of the study later published in Food Policy. The correction of the earlier "indistinguishable from zero" formulation is documented in the body. The absence of a peer-reviewed field RCT of an on-pack disposal label with objectively measured sorting is an absence claim based on the searches conducted for this article, detailed further in the final article of this theme.

Last verified: 18 August 2026.