Act for Food

Read the campaign story

Carrefour announced Act for Food on 3 September 2018 as a programme of food-related commitments supported by advertising. The release described a public website explaining actions and allowing people to express preferences about future projects. It scheduled television from 9 September, alongside other communication channels. The programme and its advertising should be understood together, without treating an announcement as completed operational change.

Contemporary production reporting credits Publicis Conseil and Marcel, with Luis Cerveró directing through Control Films. Publicis Conseil’s 2024 account describes a relaunch that placed taste and price more prominently in the proposition. This study groups that later chapter with the original family and does not independently certify the sourcing, nutritional or environmental claims made in either phase.

Editorial interpretation: a named action is easier to scrutinise than a general declaration of goodness. It creates a potential accountability surface: a customer can ask what changed, which products are covered and whether the change remains in place. That advantage disappears if the action language is specific enough to sound credible but too vague to verify. The communications team therefore needs access to operational owners and evidence, not just an attractive campaign identity.

Editorial interpretation: participation is most meaningful when its scope is clear. Voting between proposed initiatives does not mean customers control the whole business, and it should not be presented that way. An adaptation can invite input on a genuinely open decision while explaining constraints and publishing the response. This creates a smaller but more credible relationship than promising that every purchase or vote directly changes a complex food system.

Turn each promise into an inspectable statement

For every proposed action, identify the responsible team, covered products, timing, evidence and exceptions. Distinguish an ambition from work underway and from a completed change. This structure gives writers a reliable basis for concise communication and helps customers compare the promise with reality. If those details cannot be supplied, reduce the claim rather than making the language more emotionally persuasive.

Design the evidence layer before the manifesto

A memorable launch can attract questions the organisation is not prepared to answer. Build the plain-language explanation first, including limitations and an update process. Then let the campaign point toward it. This reverses the temptation to create a powerful public commitment and search for support afterward. The evidence layer should remain useful after the media burst, when customers encounter the promise in ordinary shopping.

Keep affordability and impact claims distinct

A product can be accessible in price without meeting every environmental aspiration, and a sourcing change can have costs or trade-offs. Explain these dimensions separately instead of bundling them into a single claim of being better. This allows people to make informed choices and prevents one positive attribute from implying unsupported superiority across the entire offer. It also gives the business more precise questions to investigate.

A small test for another brief

Proposed adaptation: publish a small retailer’s three documented service or sourcing changes, each with an owner, scope and dated explanation, then invite feedback on one genuinely undecided next step.

Measure whether visitors can explain one change accurately, identify its limits and find the supporting information. Track unanswered questions and correction requests as useful evidence, not as negative engagement to suppress.

Stop if a claim lacks support, implies unverified health or environmental benefits, or presents customer consultation as control the business has not granted.

• Use only changes supported by current records.
• Label proposed, underway and completed work separately.
• Publish a response to feedback without promising that a poll overrides operational constraints.