The cheapest brand research in India is your own ad account

The problem with asking people what they think

Traditional positioning research asks consumers to predict their own behaviour — and consumers are famously terrible at it. Focus groups reward articulate participants and produce consensus answers; surveys measure stated preference, not revealed preference; and both happen outside the actual moment of choice. The D2C alternative is structurally better: put competing positionings in front of real buyers as real ads, and let the checkout count the votes. Revealed preference, statistically meaningful sample, and the research pays for itself in sales while it runs.

Positioning angles, not creative variants

The discipline that makes this research and not just media buying: test strategically distinct value propositions, not colour variants of one idea. For a millet snack brand, genuine positioning angles look like: the health angle ("protein your kids will actually finish"), the tradition angle ("what your grandmother called a snack"), the convenience angle ("the 4pm drawer upgrade"), the ingredient-purity angle ("five ingredients, all pronounceable"), the taste-first angle ("junk-food taste, none of the junk"). Each implies a different brand, a different shelf strategy, a different competitor set. That is a positioning study wearing a campaign's clothes — the creative mechanics follow the standard testing structure.

A focus group costs three lakhs and tells you what eight people said. An angle test costs one lakh and tells you what eight thousand people paid.

Designing the experiment properly

The rules that keep it honest. One variable: each angle gets the same format, budget and audience — broad targeting, so the algorithm's audience-finding becomes part of the result rather than a confound. Enough budget to exit learning: ₹15–30k per angle over two to three weeks; starving a cell produces noise wearing a verdict's face. Matching landing context: send each angle to a PDP section or landing variant that continues its story — a purity angle crashing into a taste-first page measures the crash, not the angle (the scent-trail problem). And read past the click: CAC and first-order value decide, not CTR — the tradition angle may click less and buy bigger.

Reading the results like a strategist, not a media buyer

The output is richer than a winner. Read: which angle bought customers cheapest (your acquisition positioning), which brought the highest first-order value and best 60-day repeat (your brand positioning — sometimes a different answer, and that tension is strategy gold), which audiences the algorithm found for each angle (Meta's delivery is a market-structure report: the tradition angle finding 45+ metros while convenience finds 25–34 Tier 2 is segmentation data agencies would charge lakhs for), and what the post-purchase survey says buyers think they bought — the words they use become your copy bank. Repeat quarterly: positioning is not a one-time discovery but a moving target the account tracks continuously.

From test to brand: cascading the winner

A winning angle is only research until it cascades. The sequence: rewrite the homepage hero and hero-PDP around the winning promise; re-shoot the top of the creative pipeline in the winning register; align the marketplace listings' first image and title (where the same shopper meets you next); brief packaging for the next print run — the shelf claim should be the ad claim (consistency is the recall engine); and retire the losing angles from paid, but archive the learning — losing angles often win later in new segments or seasons. The whole loop, test to cascade, runs in a quarter and costs less than one focus-group cycle.

Where this replaces agencies — and where it does not

Honesty about scope: the ad account tests propositions brilliantly — which promise, which language, which emotional register moves money. It does not replace the upstream thinking: category choice, competitive frame, naming, distinctive-asset design — the strategy work that decides what to test. The best model we know pairs the two: strategic hypotheses formed properly, then settled by paid data instead of conference-room debate (this is precisely how our strategy practice runs positioning engagements — the workshop produces hypotheses, the account produces verdicts).