How one of India's largest FMCG companies tested 10 celebrity faces in under 48 hours

The question the brand needed answered

The brief sounded simple: "Out of these 10 actresses, who should be the face of our beauty campaign?" But "who is the best face" is really several questions in one:

• Who do women link with beautiful skin, glow, and natural beauty?
• Who owns beautiful hair?
• Who feels stylish, confident, and modern?
• Who do they trust and relate to?
• And when forced to pick on one word, khoobsurat, who wins?

A celebrity can be famous and still not own the one perception your brand needs. So we designed the study to measure each perception separately, instead of asking one vague "who do you like" question.

How we set up the study

1. The right women, not just any panel

We screened for women aged 22–49 who regularly use hair-care and skin-care products, the brand's actual buyers. The final sample was 118 women: Mumbai (46), Delhi (36), Lucknow (20), and Nagpur (16). Mixing metros with smaller cities mattered, because a face that works in Mumbai may not work in Lucknow. As you will see, it didn't.

2. A recognition check before any opinion

Each woman was only asked about actresses she genuinely recognised. We showed photos and asked her to name them first. This one step removes a lot of noise, because you never collect opinions about a face someone is only pretending to know.

3. Rank, don't rate

We measured 11 perceptions: skin, glow, natural beauty, hair, style, confidence, modern, trust, relatability, aspiration, and overall liking. For each one, women picked their top 5 actresses and ranked them. Ranks were weighted (first pick gets 5 points, fifth gets 1) and shown as a share of the maximum possible score. Ranking forces a real choice. Rating scales let everyone score "4 out of 5" and tell you nothing.

4. A voice "why" behind every ranking

This is where AI moderation makes the real difference. After each ranking, the AI asked the respondent why her #1 pick fits best, in English or Hindi, whichever she was comfortable in, and captured the answer in her own voice. So every number in the report came with real consumer language behind it.

5. One forced choice at the end

Finally, a fixed shortlist of 5 faces was ranked on a single word: khoobsurat. No escape, no "all of them are nice". One forced beauty verdict.

What the brand learned

To respect client confidentiality, we have anonymised the brand and the celebrities. The findings below are real; the names are coded.

There was no single winner. There were two, winning in different ways

The top two actresses finished almost level overall, but for completely different reasons. Actress A won the most perceptions and swept the entire "face" cluster: liking, skin, glow, and natural beauty. Actress B won fewer perceptions outright but never scored low on anything. She owned the identity space (modern, relatable, aspirational) and won the forced khoobsurat vote. One is a specialist, the other an all-rounder. Which one to sign depends on what the campaign needs to say.

The blind spot: the "face winner" was last on hair

Here is the finding that alone justified the study. Actress A, first on every skin and beauty perception, came last out of all 10 on hair. For a company selling both skin and hair products, signing her as one face for everything would have quietly weakened the hair story. No one in the boardroom saw this coming. The data made it impossible to miss.

Different cities crowned different queens

The national picture hid big local differences. In Lucknow, a completely different actress swept skin, glow, and natural beauty. Hair had a different winner in every single city. And the youngest women (22–30) disagreed with the national leader on almost everything. A single all-India casting call would have been strong in some markets and weak in the ones the brand most wanted to grow.

Nobody owned trust

Trust had the lowest winning score of all 11 perceptions, and the "most trusted" crown changed hands in every city and age group. The practical takeaway for the brand: whoever you sign, credibility will have to be built by the campaign. You cannot simply borrow it from a famous face.

"Khoobsurati" itself means two different things

The voice answers revealed something no ranking could. When Indian women say khoobsurat, some mean polish (skin, hair, styling, "ageless") and others mean naturalness (saadgi, simplicity, no show-off). Two meanings of beauty inside one word. That insight shaped not just the casting choice, but how the campaign itself should talk about beauty.

Why this needed AI moderation

The honest summary: this study combined the scale of a survey with the depth of qualitative interviews. That combination is painful to do the traditional way, and routine with AI moderation.

What you can copy from this study

• Break "who is the best face" into separate perceptions. Fame is not the same as owning skin, hair, or trust.
• Force rankings, not ratings. Real choices reveal real preferences.
• Always ask why, in the consumer's own language. The khoobsurati-has-two-meanings insight only existed because of voice follow-ups in Hindi.
• Cut by city and age before deciding. A national average can hide the exact markets you care about.
• Check the weakness, not just the strength. The most valuable slide in the report was the one showing where the front-runner loses.

Where InquiSight can help

If you are choosing a celebrity, an influencer, or even just a campaign direction, we can run this exact kind of study for your shortlist, with your consumers, your cities, and your language mix, and put the evidence on the table before the contract is signed. You can share your brief here or book a demo. To understand the method more deeply, read our guide on AI-moderated interviews vs traditional research agencies.