Cover at the point of need

Embedded insurance is an old idea with new plumbing. Whether it serves people depends on three things the plumbing does not provide: a real need, an underwriter who designs for it, and a willingness to be judged on claims paid.

Selling insurance at the moment someone buys something else is not new. Travel agents sold travel insurance, car dealers sold loan protection and banks sold credit insurance for decades before anyone called it “embedded”. What has changed is the technology that connects an insurer to the screen where the purchase happens, the regulation that governs what can be sold there, and the evidence about what happens when it goes wrong.

The evidence is not encouraging. In 2016 ASIC examined add-on insurance sold through Australian car dealers. Consumers had paid $1.6 billion in premiums and received $144 million in claims, a claims ratio of 9 per cent. Dealers had received $602 million in commission, with commission on some products as high as 79 per cent of the premium. Three years later ASIC found that consumer credit insurance on credit cards paid 11 cents in claims for each dollar of premium. In the United Kingdom, the FCA found guaranteed asset protection insurance paying about 6 per cent of premium in claims while commission reached 70 per cent, and in 2024 insurers covering most of that market paused sales until commission came down.

These were embedded products by any modern definition. They were offered at the point of sale, by the seller of the main product, with the insurer out of sight. They sold well because they were placed at the right moment. They failed customers because the seller’s interest was in the commission rather than the cover, and because nobody outside the chain could see the claims ratio.

That history is the right starting point for anyone now proposing to embed insurance into a marketplace, a platform or a membership organisation. The technology makes it easier to do. It does not make it more likely that the product will be worth buying.

What has changed

Three things are different in Australia in 2026.

The rules are specific. Since October 2021, insurance sold “in connection with” another product has been subject to the deferred sales model: for most add-on cover, no sale until four days after the customer commits to the main product and receives prescribed information, and then only in writing unless the customer makes contact. The design and distribution obligations require insurers to define a target market, distributors to keep sales within it, and both to monitor complaints and claims. The hawking prohibition bans offers during unsolicited real-time contact, which ASIC has made clear includes live chat and, by extension, AI assistants. Travel, motor, home and contents and several other classes are exempt from the deferral; the exemptions, due to end on 5 October 2026, have been extended to 2031. Personal accident, event cancellation and hire-related covers are not on the exempt list.

Some cover is now compulsory. On 17 August 2026 the Fair Work Commission’s interim minimum standards order for on-demand delivery workers took effect. Platforms must provide personal accident cover at “a reasonable minimum level of cover reflective of ELWs’ work on the application”. About 250,000 workers are affected. The order does not say what the minimum is. Each platform must now decide, and be able to defend its decision to workers, unions and the Commission.

The connecting technology is ordinary. Insurers and intermediaries now offer quote-and-issue interfaces, embeddable components and multi-insurer orchestration as standard. Cover Genius, founded in Sydney, reports more than 200 distribution partners and 50 insurers; Chubb Studio, launched in 2020, provides software kits that let partners show Chubb products in their own apps. A partner integration that once took a year can now take weeks. One United States sports registration platform reported integrating a refund-protection product in about two weeks with one developer.

How large is the opportunity

Estimates of the embedded insurance market vary by a factor of ten, mainly because they define “embedded” differently. BCG puts global embedded premium at about US$13 billion today, rising to more than US$70 billion by 2030. Other forecasts that include all digitally distributed partner insurance reach several hundred billion dollars. For Australia, PwC has forecast embedded premium of about A$35 billion by 2033, around 18 per cent of gross written premium.

The more useful number for an individual organisation is not the size of the market but the share of its own revenue that protection could represent, and what it would cost to deliver.

What Camplify’s accounts show

Camplify, the ASX-listed marketplace for hiring caravans and motorhomes from private owners, is the clearest public example in Australia. Protection is part of every booking. Hirers choose a level of accident excess reduction priced per day. Owners choose a tier when they list: casual, bring-your-own insurance, or a monthly membership that includes damage protection, with a lower platform fee at each higher tier.

The results are large relative to the business. In FY26, excess reduction and insurance income was $14.4 million, 36.6 per cent of total revenue of $39.2 million; in FY25 it was $16.7 million, about 40 per cent. For this marketplace, protection is not an add-on. It is more than a third of revenue.

The accounts also show the cost. In May 2025, citing constraints from its insurers, Camplify moved owner protection to MyWay Mutual, a discretionary mutual it established, backed by protected cell arrangements and excess-of-loss reinsurance. The membership margin rose and the loss ratio in the following half-year was 49.6 per cent. Over the same period Camplify restated its FY25 accounts for additional insurance costs in Europe and booked a loss-sharing contingency. Moving from buying protection to providing it raised the margin and brought insurance volatility into a marketplace’s own results.

Two lessons follow. First, protection placed well, priced per unit of use and offered in tiers can become a material line of revenue for an organisation whose core business is something else. Second, most organisations would rather not carry the risk. What Camplify appears to have wanted was an insurer that would change the product when the marketplace needed it to. That is a statement about the supply side of the market, and it is where the opportunity for a specialist programme lies.

Four tests for a programme

Embedded cover works when four conditions hold together. The need arises in the organisation’s process: a booking, a shift, a registration, a hire or an enrolment creates a risk the person cares about at that moment. The organisation controls the step where the need arises and can change it. The exposure can be counted on a natural unit, such as a day, a booking, an hour engaged or a member-month. And a claim can be decided from a short list of facts the person can supply.

Many proposed programmes fail one of these tests. Cover offered by email after the purchase fails the second. Cover priced on an organisation’s revenue rather than on the people exposed fails the third. Cover with complex exclusions that only become apparent at claim time fails the fourth, and produces the complaints that regulators then investigate. In 2022 Allianz and its travel partner pleaded guilty to making false or misleading statements about travel insurance sold through online platforms, including policies sold to people who could not claim on them. The page the customer saw did not match the wording that applied. Perspective 09 argued that the claim is where an insurance promise is delivered; in an embedded programme it is also where the programme is judged.

Why accident and health comes first

The covers that best pass all four tests are usually accident and health products: personal accident, voluntary workers, sports injury, journey cover and short-term income benefits. They pay a set amount when a defined event happens: a fracture, a hospital stay, a period off work, a permanent disability. That structure suits embedded distribution for several reasons.

The benefits can be explained in a sentence. The exposure unit is natural: participant-days for sport, hours engaged for gig work, member-months for associations. The claim depends on facts that are easy to establish. And the product already exists in group form, which means it can be included for every member of a population rather than sold one by one.

It also has constraints that need a specialist. In Australia, accident and sickness insurance cannot pay benefits for services that carry a Medicare item number, including the gap, so benefits must be designed as lump sums, income replacement and non-Medicare expenses such as physiotherapy, dental and ambulance. Accumulation matters: a stadium, a team bus, a school camp or a fleet of delivery riders in a storm can produce many claims from one event, which is why sports schemes carry event and aggregate limits. Pricing relies on hazard classes and age bands that differ widely between activities, and few programmes have enough claims history to rate on their own experience at the start.

These are the skills of accident and health underwriters at the specialty insurers. Chubb, AIG, Berkshire Hathaway Specialty Insurance and Arch all run accident and health teams in Australia, writing group personal accident for employers, universities, unions, sporting bodies and not-for-profits. Chubb is the insurer behind the injury protection Uber provides to Australian drivers and delivery people at no cost to them. New underwriting agencies have formed around experienced accident and health underwriters, including Batch Underwriting in 2024. But the agencies are built to work with brokers, and the large embedded intermediaries concentrate on travel, retail, device and ticketing. There is little in between: a programme manager that combines accident and health underwriting with the technology to run a product inside someone else’s process. That is the gap this essay is about.

Included cover before sold cover

The single most important design choice is whether the cover is included for everyone or offered for people to choose.

Included cover, funded from the organisation’s price, a levy or a membership fee, covers the whole population. The organisation holds a group policy and the people covered are beneficiaries of it. Because the person does not buy the cover, it is generally outside the deferred sales model. Group purchasing body relief lets an association or platform arrange cover for its members without its own licence, subject to conditions. Australian sport has run this way for years: Cricket’s national club programme and the cover provided to registered baseball players are both linked to player registration.

Sold cover, offered at a checkout for a separate price, is the model most people mean by “embedded”. It suits risks that vary by transaction, such as hire damage, cancellation or travel. Its take-up depends heavily on placement, price and wording, and for personal accident and similar products it falls within the deferred sales model. A government behavioural trial in 2021 found that a short add-on information statement reduced hypothetical take-up from 38 per cent to 29 per cent, which suggests that clearer information alone removes about a quarter of purchases that would otherwise be made.

For member organisations and work platforms, our view is that the included model, with an optional top-up for higher benefits, is usually the better starting point. It covers more people at a lower cost per person, it avoids most of the conduct risks of add-on selling, and it creates a population large enough to price. The delivery-worker order makes this the default for that sector. The same logic applies to volunteer organisations, where small groups report that cover is expensive and hard to obtain and where a pooled programme is an obvious answer.

Value as the design constraint

Every serious failure in add-on insurance has had the same signature: a low claims ratio, a high commission and a seller who could not see, or did not look at, either. The design and distribution obligations now require insurers to set review triggers on these measures, and ASIC’s 2023 review of target market determinations asked for claims ratios, cancellations, declined claims and complaints to be used explicitly. In 2023 ASIC issued 38 interim stop orders over pet insurance products sold through retail and health fund brands until their target market determinations were fixed.

The practical consequence is that a programme should be designed so that its value can be measured from the first policy, and reported to the organisation as well as the insurer. That means recording take-up, claims paid as a share of premium, declined claims and the reasons, time to pay, complaints, cancellations and remuneration as a share of premium, by partner and by month. If an organisation would not be comfortable showing its customers the claims ratio, the product is the wrong product.

Remuneration needs the same discipline. Market commentary puts embedded commission anywhere from 15 to 30 per cent of premium and higher for some travel products; those figures are indicative only. In a programme where the organisation is also the policyholder, remuneration matters less than it does in a sold product, because the organisation is buying the cover for its own people. In a sold product, remuneration that is large relative to claims is exactly what regulators in Australia and the United Kingdom have acted against.

What the platform has to do

The technology for embedded insurance is not mysterious, but it is different from the technology most underwriting agencies use. A broker-facing system assumes a person reviews each risk. An embedded programme issues thousands of small policies within rules an underwriter approves in advance. It needs an interface fast enough to sit inside a checkout without slowing it, and designed to hide the offer rather than block the purchase if it fails. It needs to price per unit of exposure, to accept daily enrolment feeds for included cover, to run the deferred sales timing where it applies, to generate offer wording from the configured product so the page cannot drift from the policy, and to report value measures to every party. Generating every output from one record is the argument of Perspective 03; in an embedded programme it is what keeps the offer page and the wording in step.

It also needs the same disciplines as any delegated underwriting business: authority limits, accumulation monitoring, sanctions screening, bordereaux to the insurer and reconciliation of premium. The difficulties of several well-funded embedded intermediaries show the cost of getting this wrong. Igloo in South-East Asia processes more than 100 million policies a month for partners and is still working towards break-even; wefox in Europe reached a valuation of US$4.5 billion before its carrier ran a combined ratio of 123 per cent and the group sold or closed several businesses. Reach without underwriting discipline and automated operations produces losses at scale.

Who should run the programme

There are two ways for a platform company to take part. It can supply technology to the underwriting agencies and insurers that run embedded programmes, and leave the underwriting and the relationship with the organisation to them. Or it can sit behind a programme manager of its own, with an accident and health underwriter, insurer capacity and the organisation as its client.

The first is simpler, carries less regulatory and reputational exposure, and keeps a technology company neutral between the agencies it serves. Its weakness is that the gap described above exists because existing accident and health agencies have not built for embedded distribution, and a platform waiting for them to do so may wait a long time. The second puts the underwriting, the product and the technology under one direction, which is what organisations like Camplify appear to want, but it brings licensing, capital for any profit-sharing, conduct responsibility and a potential conflict with agency clients.

Where we would start

Where the answers are good, we would start with an included accident and health product designed by an underwriter who has done it before, priced per unit of exposure, with an optional top-up and a value review at three and twelve months.

The delivery-worker order makes this urgent for one sector and sets an example for others. Embedded insurance done this way is not a new distribution channel for old products. It is group cover, designed for a specific population, delivered through the systems that population already uses, and judged on whether it pays the claims it was bought for.

Sources and basis

This essay is general commentary, not legal, financial product or insurance advice. Figures about other organisations come from the public sources below and have not been verified by those organisations. Cuttleflow has no commercial relationship with the organisations named.

1. ASIC Report 492, the sale of add-on insurance through car dealers (2016); ASIC Report 622, consumer credit insurance (2019).
2. FCA, firms recommence GAP insurance sales following FCA action (2024).
3. ASIC Regulatory Guides 275 (deferred sales model), 274 (design and distribution) and 38 (hawking).
4. Treasury, exposure draft ASIC Amendment (Deferred Sales Model Exemptions) Regulations 2026 (27 July 2026); extension to 5 October 2031 as reported.
5. Fair Work Commission, Interim On-Demand Delivery Employee-like Worker Minimum Standards Order, MS2024/3, made 11 August 2026, in effect 17 August 2026.
6. Cover Genius release, 14 July 2026; Chubb, Chubb Studio release, 2 November 2022; Vertical Insure, WrestlingIQ partner story.
7. BCG, “Embedded insurance success: get your tech stack right” (June 2025); PwC Australia, embedded insurance forecast (April 2024).
8. Camplify Holdings Limited: FY24 Annual Report; H1 FY26 Appendix 4D (26 February 2026); FY26 Appendix 4E and Annual Report (25 August 2026); ASX announcement on MyWay Mutual (May 2025). Revenue figures are from the operating segment note.
9. ASIC media release 22-144MR (Allianz and AWP, 2022); ASIC media release 23-174MR (pet insurance stop orders, 2023).
10. Gallagher, sport claims guidance; Chubb Group Personal Accident wording (2023); hyperexponential, accident and health pricing factors.
11. AIG Australia accident and health playbook (FY2025); BHSI accident and health overview, Australia (2025); Uber Australia, injury protection; ITIJ on Batch Underwriting (January 2024).
12. ASIC Corporations (Group Purchasing Bodies) Instrument 2018/751 and Regulatory Guide 195; Cricket Australia National Club Risk Protection Programme; Gallagher, Baseball Australia insurance programme.
13. Behavioural Economics Team of the Australian Government, “Slowing down to add it up” (2021); Volunteering Victoria, submission on insurance (March 2026).
14. MinterEllison, results of ASIC review of general insurance TMDs (2023); Curacel and Kasko on embedded commission (vendor sources, indicative only).
15. e27 on Igloo FY2025 results (28 September 2026); Coverager (17 May 2024) and S&P Global Market Intelligence (22 May 2024) on wefox.

Cuttleflow Systems · Perspective 20 · 33°53′S · 151°16′E · Sydney