The 90-Day Catch in India's Gig-Worker Social Security Scheme

India has done something most of the world still only talks about. With the Social Security (Central) Rules notified in May 2026, the country moved to bring gig and platform workers — the delivery riders, the app cab drivers, the people who bring your groceries in ten minutes — under a formal gig worker social security umbrella. Registration on the e-Shram portal, health cover, accident insurance, a welfare fund built partly from what the platforms themselves pay: on paper, this is one of the first serious attempts anywhere to give the on-demand workforce the kind of protection salaried employees have long taken for granted. I want to be fair about that first, because it genuinely matters. And then I want to be honest about the catch buried inside it.

The catch is a number. To qualify for the full basket of benefits, a worker reportedly has to clear a minimum-engagement threshold — described in most coverage as around 90 days on a single platform in the preceding financial year, or roughly 120 days if their work is split across several apps. That figure is as reported, not something I would state as settled law, and the exact contours are still being read out of the rules by people who do this for a living. But if it holds anywhere close to that shape, it changes the whole story. Because a benefit is only as real as the number of people who can actually reach it.

The genuine progress, stated plainly

Let me not undersell this. For years the standard defence of the gig model was that these workers were “partners,” not employees, and therefore fell into a gap where no labour protection reached them. No provident fund, no insurance, no floor under a bad month. The 2026 rules push directly against that gap. They put the obligation on the aggregators — Swiggy, Zomato, Ola, Uber, Rapido, Blinkit, Zepto and the rest — to register every worker they engage, in near real time, and to feed that data to a central portal. They route health cover and accident insurance toward those registrations. They ask platforms to contribute a slice of their turnover into a fund meant for the people generating it.

That is a real shift in who carries the burden of proof. For the first time, the default is that a rider should be covered, and the platform has to do the paperwork to make that true, rather than the worker having to fight through a system that pretends they do not exist. Set against the recent history I have written about elsewhere — the slow, patient work of gig workers unionising just to be acknowledged as workers at all — this is not a small thing. It is the state finally writing their existence into law.

A benefit is only as real as the number of people who can actually reach it — and that is exactly where the design gets interesting.

Where the 90-day threshold quietly turns the lock

Now the honest part. Read the eligibility condition carefully and you notice it does not measure need. It measures attachment to a platform. By most accounts a “day of engagement” is counted as any calendar day on which a worker logs in and earns something for an aggregator — a low bar per day, which sounds generous. But you need to stack up enough of those days, reportedly around ninety of them, inside a single financial year, with essentially the same platform, to unlock the full benefit. And here is the thing the number quietly assumes: that gig work is steady, single-platform, and continuous. For an enormous share of the people actually doing it, none of those three things is true.

Think about how this work is really lived. A rider does mornings on one food-delivery app and evenings on another because neither alone gives enough orders. A driver drives for a few intense months to cover a wedding or a medical bill, then drops off. A student rides during exam season and vanishes after. Someone moves cities, switches apps, gets deactivated by an algorithm for a fortnight over a customer complaint they never got to contest. Every one of those perfectly ordinary patterns is precisely the pattern that struggles to accumulate ninety continuous-enough days on one platform. The multi-app worker, told they instead need something like 120 days across aggregators, is being asked to hit an even higher bar for the crime of not putting all their labour into one company’s basket.

The bar is drawn where the intermittent worker can’t reach

So who clears it? The full-time, high-commitment, single-platform rider who is already the most visible face of the gig economy — and, not coincidentally, the worker the platform most wants to keep loyal. Who tends to fall short? The intermittent, the part-time, the multi-homing, the seasonal, the newly deactivated. In other words, a large chunk of the very population the scheme is named after. I would not put a hard percentage on it, because nobody honestly can yet and the data is thin. But you do not need a precise figure to see the shape of the problem: the eligibility rule rewards exactly the behaviour that makes a worker most dependent on, and most useful to, a single aggregator.

That is worth sitting with, because it is not an accident of drafting. A threshold is a design choice about who deserves protection, and a threshold set at continuous single-platform engagement is a choice that treats loyalty to a platform as a proxy for deservingness. It is the difference between a right and a reward. A right you hold because you are a worker who was hurt or fell ill. A reward you earn by logging enough qualifying days for the right company. The 2026 rules, for all their genuine ambition, lean toward the second. And there is a further sting: because eligibility resets every financial year and reportedly carries nothing forward, a worker who qualifies one year and has a lean or fragmented next year simply drops out again. Protection that evaporates the moment your work gets precarious is protection calibrated for the people who need it least.

A threshold set at continuous single-platform engagement quietly turns social security from a right you hold into a reward you earn for the right company.

Why the design looks like this — and who it suits

None of this is unique to gig work, and I do not think it is simple malice by anyone drafting the rules. Thresholds exist everywhere in social policy; you need some line to keep a scheme administrable and to keep it from being gamed. The problem is that when the line gets drawn, the people in the room drawing it have interests, and the platforms are very much in that room. A minimum-engagement bar tied to a single aggregator is, conveniently, also a retention tool: it gives every rider a reason to concentrate their hours on one app rather than spread them, and it quietly caps how many workers a platform ends up owing full benefits to. A protection that doubles as a loyalty lock-in is a protection the platforms can live with. That alignment should make us look harder, not softer, at the design.

This is the pattern I keep returning to — the way a technology or a policy built in the name of ordinary people gets subtly bent toward the institutions that mediate it. It is the same dynamic I traced in writing about how technology gets captured: the tool is real, the benefit is real, and yet the terms of access are quietly set by whoever controls the platform, so the upside pools where the power already sits. A social security scheme routed entirely through the aggregators’ own reporting, gated by days-on-platform, is not immune to that gravity. It is a live example of it. And it sits inside a larger machinery — the same gig economy exploitation that a per-day, deactivation-at-will, algorithm-managed model produces — where the worker’s instability is not a bug the platform tolerates but a feature it is built on.

What would make the promise real

I want to be clear that the answer is not to sneer at the scheme and walk away. The frame is right; the machinery is worth fixing. A few things would go a long way, and none of them are radical:

• Aggregate the days, not the loyalty. Count qualifying days across every platform a worker uses, on one e-Shram identity, rather than forcing a single-aggregator total. The worker is one person; their protection should follow the person, not the app.
• Lower the floor for the core benefits. Health and accident cover — the things that stop one bad day from becoming ruin — should attach at registration, or at a far lower day count, because those are the risks that hit the intermittent worker hardest, not least.
• Let eligibility carry, not reset. A worker who qualifies should not fall off a cliff after one fragmented year. Some carry-forward, some grace period, turns a reward back into something closer to a right.
• Audit the reporting. When the platform is both the party that owes the benefit and the party that reports the days, the incentive to under-count is obvious. The counting has to be verifiable by the worker and by the state, not taken on trust.

These are not utopian asks. They are the difference between a scheme that protects the median gig worker and one that mainly certifies the workers who were always going to be fine. And they matter more, not less, as the wider legal ground shifts: the same period has brought a broader overhaul in India’s new labour codes, and the thresholds and definitions written now will harden into the defaults millions of people live under for years.

So here is where I land. India deserves genuine credit for being early — for saying out loud that the person who brings your dinner in the rain is a worker owed something by the system that profits from them. That sentence alone puts the country ahead of many richer ones. But early is not the same as done, and a promise is not the same as a payout. The 90-day catch, if it settles anywhere near where it is reported, means the scheme could be technically historic and practically hollow for a large share of the people it names. The measure of this policy will not be how bold its announcement was. It will be a quieter number nobody has published yet: of every rider and driver the scheme claims to cover, how many ever actually clear the bar. Until that number is high, the right response is exactly this — generous about the intent, and relentless about the catch.