Show the person, the place, the capability and the evidence.

The consumer evidence points the same way, with a caveat. A 2026 Meta-commissioned Ipsos study (4,000+ respondents in 23 cities) reported that at least 97% of surveyed users in India watch video daily on Meta’s platforms. It also reported that Reels drives 81% product discovery, 66% consideration and 47% purchase influence among users, a base the release does not define^[1]. A 2025 Meta-commissioned GWI study found that nearly eight in ten Indian shoppers discover products through social media^[2]. These are commissioned studies and should be read as directional.

The story should lead straight into commerce. Imagine a 40-second video from a workshop that specialises in solid teak and custom pieces illustrative. The consumer taps through to its capability (custom furniture, teak, restoration, installation) and its evidence (verified identity, plus completed projects that each carry a verification status). The commercial information follows: typical project range, lead time, service area and current availability. From there they can request a quote, book a consultation or buy.

If the system rewards better-looking videos over better capability, it becomes another attention marketplace in which the best content creator outranks the better craftsman. Keeping video out of the ranking is necessary but not sufficient, because a video still sways the choice once it is on screen. Four mitigations follow.

1. Capture should be standardised and assisted, with the same length and structure for every provider, recorded from a template or by a field team.
2. Structured proof should appear with equal prominence beside every video.
3. Every video should be linked to a verified identity and dated, since the evidence review finds video strengthens trust only when it is tied to identity and recency.
4. The pilot should measure whether video lifts conversion without lowering outcome quality. The cost of producing and refreshing video counts as part of representation cost (Shift 2).

Suppose three providers can install a modular kitchen and all three show 4.7 stars. The rating doesn’t say whether they’ve installed kitchens before, how many, or of what type. It doesn’t say whether they’ve worked in this locality, have trained installers, or can coordinate electrical and plumbing work. It says nothing about typical completion time, whose warranty applies, or what happens when something goes wrong. One number compresses all of it.

The section drafts proposed three kinds of proof, and the evidence review adds a fourth layer, decision transparency.

These layers should not collapse into one score. A new provider can have strong identity and a thin performance history, and in some trades strong performance comes with few formal credentials. When a provider appears, the explanation should be plain: verified business identity; [n] completed appliance repairs in your area; specialises in this appliance category; available tomorrow. That says far more than 4.8 stars (Shift 5 covers why that explanation must come from the real selection logic).

ONDC’s 2022 trust consultation paper already gave a wide range of trust areas their own sections: search and discovery, order placement, fulfilment, payment and settlement, returns, refunds and cancellations, issue and grievance management, participant KYC, common taxonomy, cataloguing as a service, scoring and badging, enforcement and compliance, and community governance^[3]. ONDC’s FY 2025–26 report says it aims to enable verifiable financial identities, which is a financial credential rather than a capability one^[4]. Portable capability credentials also exist in practice: Urban Company’s service professionals receive Skill India-certified digital credentials through its partnership with NSDC^[5]. What rarely travels is performance history. The open question is how deeply trust connects to capability discovery, and whether performance evidence can become as portable as certificates.

Cold start. New providers have no transaction evidence, yet they need demand to build it, and a trust layer that rewards history can lock in incumbents. Several options are worth testing: verified identity plus capability proof as sufficient for entry; a visible “new” label instead of exclusion; rotating ties in shortlists (Shift 5); a user choice on whether to include new providers; and concierge follow-up and verification on early jobs.

Verification cost. Someone has to check licences, portfolios and claimed job counts, and human verification may be one of the largest operating costs in the model. It can be tiered, starting with automated identity checks, then sampled field verification, with transaction-derived evidence as the cheapest source over time. Whatever it costs belongs in mission completion cost (Shift 1).

Gaming. Self-dealing transactions and bought reviews can manufacture evidence, and portable reputation raises the payoff for gaming. Transaction evidence can be weighted by the diversity of counterparties, a sample of completions can be verified, and dispute history can be shown.

Portability. Portable trust depends on network governance, and ONDC says only that it aims to enable verifiable financial identities^[4]. For now portability is a direction, not a capability.