It should feel like entering a market.
FromResults grid→ToEntering a market
A search result is an answer. Discovery is an experience. Short, human introductions bring back what the grid left out: people, craft and the chance of a surprise, while the agent keeps the set relevant and the buyer keeps control of what happens next.
A market on a phone
Walk through a great physical market. You don’t inspect every stall. Something catches your eye, you stop, and someone tells you why their work is different. Sometimes you leave with something you never came for.
Now picture the same thing on a phone. The screen says: “I found 18 people who fit what you’re looking for.” Then, one at a time illustrative:
“Hi, I’m Neha. I make small-batch Japanese-style cakes from my home kitchen in Rajpur…” Swipe.
“I’m Ishaan. I turn traditional Uttarakhand flavours into modern celebration cakes…” Pause.
“We’re a three-generation bakery…” Save.
Digital marketplaces replaced the physical market with search, filters and grids, making commerce fast, convenient and comparable for millions. This shift brings back what the grid left out: people, stories, craft and serendipity.
The Discovery Reel
The listing tells me what you sell. The reel lets me discover why I might want to meet you.
Fifteen seconds, four questions: who are you, what do you do, what makes you different, and why should I care?
A phone camera is enough. The real workshop, the real kitchen, the real person. A seller can have many reels: who we are, what we make, how we make it, what’s new this month, what a customer experienced. The same bakery can lead with celebration cakes for one buyer and low-sugar desserts for another.
Precision first, then serendipity
The agent draws a boundary around what’s relevant. Inside it, the buyer wanders. AI should reduce the search space without reducing the possibility space.
Designed diversity. A discovery set might hold, for example, eight close matches, three adjacent possibilities, two unusual options and two emerging participants illustrative, so newcomers have a fair chance to be seen when they’re relevant.
Depth on demand. Reel, then offering, evidence, reputation, conditions, conversation and transaction. Information appears as interest grows.
Every interaction is a signal. A pause, a rewatch, a save or a question refines the search. A swipe isn’t a yes or a no.
Comparison comes later, once the buyer knows which differences matter.
Go deeper
Why video
Short video is already a major way people in India find products and businesses. Meta reports, from a study it commissioned from IPSOS with more than 4,000 respondents in 23 cities, that 84% of Gen Z consumers and 73% of rural users discover new products and brands on its platforms[1]. An earlier Meta-commissioned study by GWI found that nearly 8 in 10 shoppers discover new products through social media[2].
Scope. Both studies were commissioned by the platform, and neither states every base, so we treat them as directional. They show that video is where many people already look. They don’t show that it finds better matches, which is what the test is for.
Discovery, not addiction
The goal isn’t time spent in the feed. It is the quality of possibilities encountered per unit of attention. The feed ends when the relevant options do. Success is measured by relevant encounters, meaningful conversations and successful outcomes, and by how often small and new sellers are discovered.
Getting it right
Real sellers, real products. AI can help with scripts, captions, trimming and translation, but not with synthetic presenters or invented products. AI assists representation. It shouldn’t manufacture authenticity.
Signals serve the buyer only. Behaviour is used to improve that person’s own discovery, and is never sold or used for ad targeting. Sellers learn in aggregate what interests buyers, never who those buyers are.
Sponsored content is always labelled. Money never quietly becomes relevance. India’s consumer protection authority names “disguised advertisement” as a dark pattern in its 2023 guidelines, which apply to platforms, sellers and advertisers[3], and ONDC’s 2022 consultation paper proposed that boosted or sponsored results be clearly marked[4]. Shift 7 returns to how.
What could go wrong
- Reels become a chore.If making a video is hard, small sellers are left out again. A phone camera has to be enough, and help with scripts and captions has to feel optional.
- Performance over substance.Polished presenters could win over good makers, so evidence sits one tap behind every reel. Hyperlocal, Shift 6 looks at narrative and proof.
- Attention capture.A feed can drift towards time spent. The measure is outcomes, and the feed ends when the relevant options do.
- The same faces again.If a few sellers’ reels dominate, designed diversity and a place for newcomers have to be tested, not assumed.
How we’d test it
Reels for 100 sellers in one category, with a discovery feed for a group of real buyers.
- The test
- Reels against a grid
- What we would measure
- Compare the feed with a traditional listing grid. Measure discovery quality, how often small sellers are discovered, successful outcomes, and whether people feel the time was well spent.
- What would end the idea
- Reels don’t beat a listing grid on outcomes, or sellers stop making them.
Sources
- [1]Meta, Reels is shaping India’s video-first future across Gen Z, women and Bharat (3 June 2026)Checked; a company announcement of a study it commissioned from IPSOS (more than 4,000 respondents, 23 cities). Base populations are not stated, so directional
- [2]Meta, Creators, short-form videos and messaging are disrupting how offline retailers sell and scale in India (12 June 2025)Checked; a company announcement of a study it commissioned from GWI. Sample size is not stated, so directional
- [3]JSA, CCPA issues guidelines for prevention and regulation of dark patterns, 2023 (guidelines notified 30 November 2023)Checked via a law-firm summary; the Guidelines’ own text not read
- [4]ONDC, Building Trust in the ONDC Network: Consultation Paper (29 September 2022), section 4.1.1Checked; proposals for comment, not adopted policy
Spotted a source we missed or a rule we misread? Tell us what we got wrong.
Meeting a seller creates interest. Confidence needs something firmer: what makes this seller distinct, and what backs it up.