This piece is by Dan Bratshpis, CEO and co-founder of INSHUR
The insurance industry has spent the last decade congratulating itself for discovering the gig economy. The self-congratulation is premature.
When platforms like Uber, DoorDash, and Amazon found that conventional insurers couldn’t serve their drivers, they did what tech companies do: they built around the problem. They absorbed the risk, embedded coverage into their apps, and got drivers on the road. From the outside, it looked like a coverage gap closing. In practice, it was the industry’s original failure being handed off to someone else to manage.
Reluctant insurers
No platform set out to be in the insurance business. How they ended up there differs by market, but the destination is broadly the same.
In the US, Uber, Lyft, and DoorDash became major insurers by choice as much as by necessity. The conventional market couldn’t produce products that worked for drivers logging unpredictable hours across multiple platforms, so platforms absorbed the risk themselves. What started as a practical workaround became competitive practice: once one platform covered its drivers, others followed, and blanket coverage across the entire driver pool became the default.
In the UK, the framework points toward driver-owned insurance as the more appropriate model. Hire-and-reward coverage is the driver’s responsibility, and the market has been moving in that direction; Amazon Flex was one of the last platforms providing blanket cover and has since stepped back from it, providing driver incentives instead. The result is a workforce pointed toward an open market selling products designed for full-time taxi operators and couriers, not someone delivering parcels three evenings a week to supplement their income.
In the EU, platforms are increasingly classified as employers by regulation, which places coverage obligations on them directly and recreates the blanket pooling dynamic seen in the US, just through a different mechanism. One response that makes sense given those constraints is co-coverage: platforms carry the base layer of risk, with third-party insurers covering the additional exposure. Where that third-party layer is priced on individual driver data rather than pooled across the platform, it reopens the door to the kind of differentiated pricing the blanket model forecloses.
The model has merit beyond the EU regulatory context. Co-coverage offers a route to transfer additional or on-demand risk to insurers equipped to price it properly, without disrupting the driver experience platforms have built around it.
In each market the structural problem is the same, even if the route into it is different.
The problem that replaced the problem
Embedding coverage into platform apps solved the friction problem. Drivers got coverage without navigating a broker or deciphering a policy document, and take-up rates reflected that. But removing the friction didn’t fix the underlying product, and the coverage most drivers receive today is still fundamentally broken in the same way it always was.
Platform-based insurance is generally priced across their entire driver pool, mutualising risk rather than differentiating it. Coverage isn’t priced on the individual driver; it’s priced on the platform. The reason this persists isn’t that individual pricing is technically impossible. It’s actually straightforward: gig drivers are the primary focus for these platforms. Without drivers, the platform doesn’t function, so platforms will do anything that reduces friction in getting drivers signed up. Insurance was a friction problem before it was a pricing problem, and platforms solved the friction without ever addressing the pricing, and the path of least resistance hardened into industry standard. What passes for innovation in this space is largely a convenience wrapper around the same broken underwriting the conventional market was criticised for, now absorbed into platform economics and rebranded as a feature.
The people absorbing the real cost of this are the better drivers. Anyone who drives carefully, maintains a high platform rating, and takes fewer risks is subsidising colleagues with worse records who are pooled alongside them. Most drivers have no idea this is happening, because the coverage is embedded and the pricing rationale is never surfaced.
What the data actually shows
The frustrating part is that the data to price these risks individually already exists and is already being generated.
Our own data makes this concrete. Among drivers that any underwriter would classify as top tier, the difference between the best and the very best represents over a 20% difference in accident frequency. Conventional pricing treats them as the same risk. They aren’t. That level of differentiation is only visible if you are actually pricing on the driver, and most of the market isn’t.
Traditional pricing variables still matter. But in an era where platforms generate real-time data on every driver, every shift, every trip, the question isn’t which data source to trust: it’s why you wouldn’t use all of them. The tools to process that complexity, AI, big data, modern actuarial modelling, exist and are accessible. What’s missing is the urgency to build products that require them.
What good looks like
Insurance that actually works for gig drivers prices on the driver rather than the pool. It draws on platform-generated data (ratings, driving behaviour, hours worked) to assess individual risk, rather than falling back on demographic proxies that have no relationship to how a specific person actually drives. It’s active when a driver is working and reflects real time on the road, rather than spreading an annual premium across twelve months regardless of how much driving gets done.
Where individually-priced products exist, the results are broadly consistent: better driver retention on platform, stronger loss ratios for carriers, and coverage that drivers can make sense of because the pricing reflects their own behaviour. Removing purchase friction and maintaining pricing rigour are not in conflict with each other, though the industry has often treated them as though they are.
A concrete ask
Regulators need licensing frameworks designed to accommodate individually-priced, usage-based products rather than forcing them into annual policy structures they were never built for. In the US, fifty separate state filing requirements each adding their own friction doesn’t produce well-designed gig economy products. It produces blanket coverage, because that’s what clears approval fastest.
Platforms should also be asking what carrying insurance risk is actually costing them. Holding insurance obligations requires regulatory capital, capital that sits in reserves rather than being deployed in the technology, operations, and growth that are their actual competitive advantage. The platforms absorbing the most risk aren’t just solving a coverage problem on behalf of their drivers; they are running an insurance business inside a logistics or mobility business, and those two things have very different capital requirements.
Insurers (particularly those who ceded this market to platforms in the first place) need to compete on the quality of their underwriting, not just the convenience of their distribution. Wrapping a conventional product in a gig platform app is not a meaningful answer to what this workforce actually needs.
Who owns this market in ten years
The gig economy is not going to become simpler to insure. The workforce is growing, risk profiles are fragmenting across more platform types, and the regulatory environment in most markets is not moving in a direction that makes this easier. The insurers who invest now in individual risk pricing, built on real behavioural data, in products that can operate at the speed these platforms require, will be the ones setting the terms of this market over the next decade. The ones still writing blanket policies and describing it as embedded insurance will find themselves subsidising someone else’s better product.
Be the first to comment