Seen, With No Measurable Cost
A meta-analysis of 155 datasets finds that ethnic minority endorsers win over minority consumers, with no measurable cost on majority consumers' attitudes or buying intentions, and that among the casting choices tested, fame was the one that weakened the effect for every audience.

A meta-analysis of 155 datasets finds that ethnic minority endorsers win over minority consumers, with no measurable cost on majority consumers’ attitudes or buying intentions, and that among the casting choices tested, fame was the one that weakened the effect for every audience.
Picture a casting meeting a few weeks before a shoot, in a room with a wall of headshots, a brief that says “relatable” and at least one person who has been asked by someone senior whether the core customer will feel the ad is still for them. The question is rarely put that bluntly, but anyone who has sat in that room will recognise its shape, because it rests on an old fear in advertising, which is that casting an ethnic minority endorser wins one audience and quietly costs you another.
That fear has usually been argued from instinct, from a single campaign that went well or badly, or from the comment section under a post, and so a large body of evidence on the question is useful, because it points fairly clearly in one direction. The answer is not quite the one either side of the argument tends to expect.
What the evidence says
Martin Eisend of Copenhagen Business School and the University of Vienna, with Anna Rößner and Erik Hermann of European University Viadrina, published a meta-analysis in the Journal of Marketing that pooled the English-language experiments they could find comparing ads fronted by an ethnic minority endorser with the same kind of ad fronted by an ethnic majority one. The paper appeared online in December 2025, and Bruno Schivinski’s practitioner summary on 7 November brought it to a wider marketing audience. It draws on 106 papers published from 1969 onwards, covering 155 unique datasets and 1,276 effect sizes from 21 countries, although most of the samples come from the United States.
Ethnic minority consumers responded more favourably to minority endorsers on a range of the outcomes measured, and according to Schivinski’s summary that included identifying with the endorser, feeling that the ad was meant for them and intending to buy. Among ethnic majority consumers, attitudes to the ad, the brand and the product, and buying intentions, showed no significant change, which the authors read as reassurance. A null result is not proof of zero, though, and the picture is not entirely clean, because in the two experiments that asked majority participants to write down the thoughts an ad brought to mind, they listed more negative thoughts when the endorser came from an ethnic minority. The authors also describe the overall effects as small, so inclusive casting, on this evidence, is a targeting gain concentrated where people recognise themselves rather than a growth lever for everyone.
The twist is about fame
The most useful finding for anyone briefing a casting agency concerns celebrities, who featured in only 65 of the 1,019 persuasion results, and yet among the casting choices tested, celebrity status was the one that weakened the effect for every audience, turning it slightly negative. The authors’ explanation is that a famous ethnic minority endorser tends to be seen as an exception rather than a typical member of their group, so viewers identify with them less. Fame is not the only thing that moved the result, because the country’s migration and legal climate, the era of the study, whether the brand was real or invented and the endorser’s gender all made a difference too, but it is the one a marketer most often reaches for on purpose.
Psychologists have a name for treating a person who stands out as a special case rather than as evidence about a group, which is subtyping, and Reneé Weber and Jennifer Crocker’s work in 1983 described it as one of the ways stereotypes survive contact with people who do not fit them. The link to this meta-analysis is my own reading, not the authors’. Social Identity Theory offers one way to read the rest of the result, which the meta-analysis did not itself test, and Henri Tajfel and John Turner argued that people derive part of their sense of self from the groups they belong to, so they respond to cues that signal this is for people like me. In Dear Algorithm I wrote about the quiet message behind a simple recommendation feature, that “other people, people like you, chose this thing”, and an endorser who looks like your neighbour may send the same signal far more directly than a famous face can.
The limits deserve stating plainly, and the first is that only 51 of the 1,019 persuasion results, about 5%, came from ads shown online, which is a thin base for the medium where most casting now runs. Most results, 725 of them, came from invented brands rather than familiar ones, the evidence leans towards Western countries, and ethnicity was mostly signalled through appearance alone.
Identity is one layer, needs are another
What that evidence opens up, in my view, is a different way of thinking about segmentation. The STAR Framework reads decisions through four situational lenses, Socialiser, Thinker, Adventurer and Realist, and this section is my interpretation, which the meta-analysis did not test. The Socialiser lens is about how much other people and shared connection weigh in someone’s decisions, and in my reading it is the closest of the four to the mechanism this research describes, because feeling seen is first a question of whether the person on screen could be someone like you.
The point I would draw out for segmentation is that identity and needs are two different layers. Identity tells you who might recognise themselves in an ad, and the lenses suggest what that person is looking for at the moment of decision, whether that is shared connection, a reason they can weigh, something new, or security, stability and reliability, along with simplicity and ease. Two viewers who share an identity can be looking through different lenses, and one viewer can switch lenses between a Saturday browse and a Tuesday purchase, so casting that earns recognition still has to be paired with a promise that meets the need in play. In my reading, good targeting uses both layers, and treating either as the whole story is how brands end up with ads that are inclusive and still not very persuasive, or persuasive to fewer people than they could have been.
A toolkit for casting with evidence
- Cast for recognition, not fame. Choose endorsers who could plausibly be your customers, because the evidence suggests the effect comes from people seeing someone like themselves, and celebrity status weakened it.
- Stop assuming a majority penalty, and test for one. On attitudes and buying intentions the pooled evidence shows no measurable cost among majority consumers, but the thought-listing results and the differences between countries are reasons to test rather than assume.
- Layer needs on top of identity. On my reading, once the casting earns recognition, the message should also speak to more than one lens, so that the ad gives a reason, an easy next step or a sense of discovery as well as a sense of belonging.
- Test online, because the research mostly did not. Run casting versions side by side in your own channels, since only about 5% of the evidence came from online ads.
- Plan market by market. The effects varied with country conditions and over time, so read results by market rather than in one global average.
The test question for the next casting meeting is short enough to write on the whiteboard next to the headshots: will the people we want to reach recognise themselves here, and will they find what they came for?
The wall of headshots will still be there next time, and the nervous question about the core customer will probably still be asked. The difference is that there is now a clearer answer to it, which is that across a large evidence base, helping one audience feel seen showed no measurable cost on how the majority rated the ad or the brand, as long as you resist the temptation to make the person on the poster famous.
If you would like to see which needs tend to shape your own responses to ads, the STAR Cognitive Profile Assessment at starassess.com takes 33 questions, and the assessment and its full report are both free.
David Chadderton is a behavioural strategist, Chief Marketing Officer and STAR Dynamics. He writes The Unoptimised Human.
References
Chadderton, D. (2026). Dear Algorithm, It’s Not Me, It’s You: Putting the Personal Back Into Personalisation (2nd ed.).
Eisend, M., Rößner, A., & Hermann, E. (2025). Ethnic diversity in advertising: A meta-analysis. Journal of Marketing. Advance online publication.
Schivinski, B. (2026, 7 November). Does ethnic diversity in advertising work, and with which audiences? What a Journal of Marketing study found. Bruno Schivinski.
Tajfel, H., & Turner, J. C. (1979). An integrative theory of intergroup conflict. In W. G. Austin & S. Worchel (Eds.), The Social Psychology of Intergroup Relations (pp. 33–47). Brooks/Cole.
Weber, R., & Crocker, J. (1983). Cognitive processes in the revision of stereotypic beliefs. Journal of Personality and Social Psychology, 45(5), 961–977.
Eight minutes. Thirty-three questions.
A precise picture of how you operate: motivations, decision style, and blind spots. Free to start.
- S
- T
- A
- R
The canonical source.
Essays, when there is something worth sending. Choose the subjects you want. I will ask you to confirm the address before I write.