How these assessments are produced

Data: the Australian Government's PrivateHealth.gov.au releases. Member benefit notes: insurers' own published pages, checked and linked as evidence.

ClearChoiceHealth is a free, non-commercial comparison guide: answer a short questionnaire and every private health insurance product open for sale in your state is filtered against your answers, with each surviving product given a market-relative value assessment. This page explains how those assessments are produced, whether you have used the tool yet or not.

1. The one question we assess

For every private health insurance product open for sale in Australia, we ask a single question:

What would the market normally charge for this product's cover, and is its actual premium above or below that?

A product priced well below what the market normally charges for equivalent cover represents stronger value; well above, weaker value. This is deliberately different from "cheapest first": a bare low price with thin cover earns no credit here. It is also not advice: the assessment says nothing about whether a product suits you, your health, or your budget.

2. A worked example

Take a fictional but typical product (the same steps run for every real one):

"Example Silver Plus Hospital $500 excess", NSW, single person, $234.00/month.

Step 1: Your requirements filter the market

If you told us a hospital category is a must-have (say, pregnancy and birth), every product that does not provide it is removed before any assessment happens. Some categories arrive already ticked, because they are commonly used by people in a similar position to yours. A tick you leave in place counts the same as one you added, so it filters too, and you can untick any of them before you continue. The results page always tells you how many products each requirement removed, and where we can work out a price, what the cheapest of them would have cost. A category covered on a restricted basis counts for most requirements; hospital psychiatric cover is the exception: the questionnaire promises full inclusion, so restricted cover does not pass that one.

Step 2: The market's expected price for this cover

From the published data, we group products by what they actually provide: the full 38-category hospital treatment profile, excess and co-payment arrangements, who is covered, and the state. Looking at how the whole market prices each cover profile, we estimate what this product's cover would normally cost: for our example, say $248/month, with an expected range reflecting normal market variation.

Every estimate carries a confidence level (high, moderate, low, or insufficient) that depends on how much comparable market evidence exists: a profile sold by many funds gives a confident estimate; a profile sold by one fund alone cannot be checked against anyone else and is treated accordingly.

Step 3: Actual vs expected

The product charges $234.00; the market-expected price is $248.00. The premium sits about $168/yr below the market-expected price (−5.6%).

Step 4: Is that difference meaningful?

A difference only counts when it is larger than the model's own measured error for predictions at that confidence level, and never smaller than normal market price variation. Small gaps land as "in line with the market-expected price"; clear gaps as "below" or "above"; only differences that clear a much higher bar earn "materially". Our example lands in "Below the market-expected price" (stronger value).

Measured error comes in two grains. Usually it is our error across every prediction at that confidence level. Where we have gone back and checked our predictions against real premiums for one particular kind of cover, we use that cover's own measured error instead. It can be larger than the general figure or smaller, and where it is larger, those products have to sit further from the expected price before we will call the difference material.

One limit of that error figure is worth knowing. When we check a prediction, the set of comparable products we check it against normally includes other cover sold by the same insurer. Insurers price their own range consistently, so part of what looks like accuracy is really an insurer agreeing with itself. So we measure our error both ways, once with that insurer's own products counted and once with them removed, and we track the gap between the two. On hospital cover the two figures differ enough to matter. Measuring it the stricter way can only reduce the number of strong claims we make, never add to them.

Where the expected price comes from is worth stating plainly. We learn it from every product in the public dataset, not only the ones currently for sale. That includes cover closed to new members and cover restricted to a particular group or employer. Those premiums are what real members pay, so leaving them out would describe a narrower market than the one Australians are actually in. Insurers tend to price the products they are currently marketing keenly, so a benchmark built only from the shop window would flatter the industry.

We only ever show you cover that is open for sale, and only from funds you have told us you are eligible to join. Learning from the whole market and showing you part of it are separate things, and that is deliberate.

One limit is worth naming. Each product counts once, however many people hold it. The public dataset publishes premiums but not membership numbers, so a legacy policy held by a handful of people carries the same weight as one held by hundreds of thousands. Read the expected price as what the market charges for this cover across every product on the books, rather than as an average of what Australians actually pay.

Step 5: Honesty gates can override the arithmetic

• Several different situations → no verdict. A product can end up "not assessed" for reasons that have nothing to do with how much evidence we hold. Sometimes we genuinely have too little to compare against. Sometimes the product has no published premium at all, so there is no price to assess. Sometimes a state government scheme already covers the service, so a market comparison would mean nothing. And sometimes we hold plenty of evidence and are withholding a verdict anyway, under the reliability rules below. The label tells you we are not making a claim. It does not tell you which of those applies, so where we can name the reason, we show it on the product itself. Either way the product appears with its factual cover details, never hidden, never ranked.
• Low confidence, accuracy unchecked → no strong claims. A "materially" verdict needs a gap wider than our own measured error for cover like this one. Where the market gives us little to compare against and we have not checked our accuracy against real premiums for that kind of cover, no gap is wide enough to earn one.
• Prices driven by features we cannot see → no negative verdict. Some products are priced on features that exist only in free-text descriptions. Calling one "weaker value" would be an accusation the data cannot support, so we say exactly that instead.
• Hospital and extras stay separate. A combined product shows each component assessed on its own; a strong overall verdict can never hide a weak component.

One consequence of the third rule is worth naming, because that rule only runs in one direction. Where a premium sits above our expected price and the product is priced on features we cannot see, we withhold the verdict. Where a premium sits below our expected price in the same circumstances, the verdict stands. So a product showing no value assessment for this reason is more likely to be priced above our expected price than below it, and reading that silence as neutral would be a mistake. The cover details we can verify are still shown in full.

3. What the premium figures mean

Every product's price starts from the fund's published monthly premium (per policy, for the stated household type and state). If your answers let us place you in a rebate tier, the headline figure is an estimated price with the Australian Government Rebate applied; the approved premium itself is always shown in the product's price breakdown, and if we can't estimate your rebate, prices are shown exactly as published. Your payable price may still differ with your confirmed rebate, Lifetime Health Cover loading and any fund discounts.

The rebate line. If you answered the income and age questions, the results page states which rebate tier your answers place you in, using the statutory income thresholds for the rates year shown, and applies that tier's statutory percentage to estimate your price. The itemised breakdown on every product shows exactly what was applied. Because your tier comes from banded answers it is always an estimate, never a quote. Two things never change with it: value assessments are computed on the approved premium, and results are never re-ordered by your personal adjustments. For household policies the tier also depends on the age of the oldest person covered, which we don't ask; and if you preferred not to answer, we simply say a rebate may apply. The same applies to the Lifetime Health Cover and age-based discount notes: both are honest pointers to rules that depend on details no questionnaire answer can compute (certified age at entry, and the discount schedule a fund applies within the 18 to 29 range, which the public dataset does not publish). Whether a product offers an age-based discount at all, and whether it carries over a discount you already receive elsewhere, are published, so we show both against each product. Confirm the exact figures on PrivateHealth.gov.au.

The Medicare Levy Surcharge note. If your income tier is at or above the first surcharge threshold, the results page notes that the surcharge is removed only by private hospital cover whose excess is within the statutory cap (currently $750 or less for a single, or $1,500 or less for a couple, family or single-parent policy) and never by extras-only cover. A cheaper policy with a higher excess does not remove the surcharge. If you told us you're young and mainly want the cheapest cover that just covers the basics, an optional filter lets you narrow your results to Basic and Basic Plus policies, the cheapest tier, which still carries the things a younger person is more likely to use (such as joint reconstructions and tonsils, adenoids and grommets). The filter hides higher tiers that cover more than you asked about, and for a higher earner Basic and Basic Plus are also the policies that satisfy the surcharge. It only hides products, it never re-orders them, and you can switch it off. If you have told us you need cover for something specific, it stays off, because narrowing to the cheapest tier would work against what you asked for. This is a pointer to a tax rule, not tax advice, and prices shown here never include or offset the surcharge itself. Confirm your own position on PrivateHealth.gov.au or with the ATO.

Provider networks are not assessed. The published data set does not include which hospitals or doctors have agreements with each insurer, so no result here reflects whether your preferred hospital participates in a product's network. This limitation applies to every product and every user equally. Check agreement hospitals for any product on PrivateHealth.gov.au before acting on a comparison.

The same cover can carry different prices. Some insurers sell identical hospital cover under more than one brand name, or as scheme variants for particular employers and groups. In the published data these are separate products with the same clinical cover, in the same state, for the same household type, at different prices. Our assessment gives no weight to the name on the box: every product is assessed on its own published premium against the market's expected price for its cover. Where a brand charges more for identical cover, that product simply reads as weaker value here. You can check any product's underwriter on PrivateHealth.gov.au.

4. How matching against your answers works

Alongside its market-value assessment, each product family shows a "Your stated requirements" note. This is not a second score. It is a restatement of your own answers against the product's published cover list:

• Only your ratings drive it. If you rated an extras category as important (the top of the importance scale), we check whether the product's published benefits include it; if you rated it not important, we say when you'd be covered for it anyway. The scale's middle option ("moderately important") is treated as neutral: moderate interest is not a stated requirement, so we never build a claim on it. Categories you were never asked about are never assumed.
• The questionnaire's categories map to the government's. "Allied health" covers the services the question named: chiropractic, remedial massage and podiatry; "major dental" includes root canal (endodontic) cover. Fit notes always name the specific services a product actually includes.
• Matching never changes the value assessment. A product's market-value band is computed from market pricing alone. The two can disagree: a product can be strong value but miss something you rated important, or match everything you rated important while being priced above the market expectation. Both facts are shown; neither overrides the other.
• No ranking. The "closest matches only" control is a filter you switch on. Like the requirements filter in step 1, it narrows what is shown and tells you by how much. Nothing is ever ordered by fit, scored, or picked for you.
• Coverage, not benefit amounts. Fit notes state whether a category is included in the product's published benefits. They make no claim about how generous those benefits are, and your spend answers are never turned into payout estimates.
• Shared annual limits. Most policies do not give each service its own pot of money: they group several services under one shared annual limit. Where a category you rated as important shares its limit with other services, we say so and name them. This is published information: we simply read the grouping the insurer declared, so you can tell apart two products whose cover lists look identical.
• Two numbers, stated separately. Where a product publishes a plain per-person annual figure for that shared limit, we show it, and where the amounts you told us you spend fall clearly above or below it, we say that too. These are two different quantities (what you spend out of pocket is not what a fund pays back), so we never combine them into an estimate of what you would get back or of whether you would be "covered".
• When we stay silent. Some products describe their limits in ways that cannot honestly be compared to a single dollar figure: limits set per policy rather than per person, limits written in free text, sub-limits inside a shared pot, or "choose your own" extras where the active services are not published. In those cases we describe the grouping and stop there. Silence means the published data does not support the statement, never that the product is worse.

"People in your position" notes. Where your journey includes hospital cover, each product may show which hospital categories it does not fully cover among those most commonly used by people in your demographic position (your age, and your household and sex if you told us). These are population service-usage rates from national Medicare data (the year to July 2016), never a prediction about you: a rate describes how often people in a group use a service, not whether you will. The notes name gaps only: they never score or rank products, and they never change the market-value assessment. Categories you already told us about are left out (you've answered those directly), and sensitive categories are never inferred from your demographics.

Where this shows up. The same usage data drives the notes on individual products. Where a product covers none of those commonly used categories in full, its card says so and names what is missing. Nothing is grouped, moved, hidden or scored on that basis: every product keeps its place in your results and the same market-value assessment, and the note ignores price and value band entirely, so two products with the same hospital cover always carry the same note whatever they charge.

Two limits worth knowing, because a number can hide both. Some categories cannot be measured from Medicare data at all: blood and adult dental are the clearest cases, and podiatric surgery has no Medicare items by regulation. We leave those out of the comparison and say so, because "we can't see it" is not "people don't need it". And Medicare renumbers its item codes: where a procedure was renumbered after 2016 we reconstruct its usage from the older codes, so the figure is a reasonable reconstruction of 2016 practice rather than a measurement of today's. Back, neck and spine surgery is the category most affected by that.

Price proximity in the shortlist sections

The sections at the top of your results are a shortlist, not the whole market. To keep them honest, each section shows only cover priced reasonably close to a benchmark, and that benchmark is the cheapest option that matches your answers among the insurers that section covers, however that cover is packaged. For hospital cover, price-framed sections stop at 1.5 times the benchmark. Value-framed sections stop at 2 times it, since a stronger value assessment can justify a somewhat higher price.

Extras cover works differently, because the cheapest extras policies on sale pay only for ambulance, or pay no benefits at all, for a few dollars a month. Measured against one of those, almost nothing else looks close in price, and the shortlist filled up with policies that cover no extras services. So on an extras journey a policy covering only a single service cannot set the benchmark, and ambulance-only and no-benefit policies are left out of the shortlist sections. Extras premiums also spread much further above their cheapest point than hospital premiums do, so the limits are wider: price-framed sections stop at 3.5 times the benchmark and value-framed sections stop at 4.5 times it.

These multiples are our editorial judgment, not a number derived from the data. We chose them so a list that starts near your price level cannot drift into products costing several times more, while still leaving each section enough cover to fill the rows it shows. When little else exists near that price level, the section says so and stays short rather than padding itself with more expensive cover. The full list further down always contains every product that matches your answers, at every price, including the ambulance-only and no-benefit extras policies that the shortlist leaves out.

Hospital plus extras results are one row per fund

Most large funds sell hospital cover and extras cover as separate policies that people buy together, while many smaller funds publish ready-made packages. When you ask about hospital and extras cover, we show one result per fund: the fund’s hospital policy and extras policy that meet your stated needs, priced as the sum of the two published premiums, alongside any packaged policy the fund sells. Whichever is better on the page’s own ordering (value group first, then price) fronts the row; the alternatives stay one click away inside it. A package is never a requirement for a fund to appear.

How the two policies are chosen is mechanical selection, not advice: the hospital policy is the one that meets your stated hospital needs in the strongest value group (cheapest first within a group), and the extras policy is chosen the same way from the fund’s extras range. Where you stated no extras preferences, ambulance-only and no-benefit policies are never chosen automatically; they remain selectable in the row. Each policy keeps its own market-value assessment: the two assessments are shown side by side and are never combined into a new one.

A summed figure is always labelled as two policies priced together. For most funds a package costs exactly the sum of its two component policies, so neither form is systematically cheaper; where a fund genuinely discounts its package, the package fronts on price like any other result.

5. Member benefits published by insurers

Some insurers publish member benefits that sit outside the official product data: medical gap schemes, no-gap dental check-ups, accident protections, rules that carry unused extras limits forward, and member programs. The government dataset this site is built on does not carry them, so we research them separately, from each insurer's own public website.

For a benefit to appear in results it has to clear each of these steps. We locate the claim on the insurer's own current page, never a third party's. Where the conditions live in a separate document, we locate that too. We then record a dated verification. From then on, the benefit appears here only while a dated check no more than 45 days old confirms the insurer still publishes the claim. Those checks are made two ways: by us, as recorded verifications, and by an automated check that runs daily, revisiting insurer pages and confirming the claim's exact wording is still there, while flagging anything that changed for us to review. If a change or an outage stops those checks succeeding, the benefit drops out once its last dated check is more than 45 days old, and stays out until it is verified again. The benefit is shown with its conditions attached and a dated link to the insurer's page, so you can read the source yourself.

When a check lapses or the insurer's page changes, the benefit is removed from results until it is verified again. A benefit you saw last month may be absent today because our verification lapsed or the page moved, not necessarily because the benefit ended. A gap can also mean a benefit reached its own stated end date, or that we retired an entry, or that the product it applied to is closed to new members. We do not claim to know why a benefit is not listed. What we know is what a page says on the day we check it, so we describe our own checks rather than the decisions of any insurer.

These checks confirm that the insurer currently publishes the claim; they are not a promise that a benefit applies to a particular product or person. Eligibility, provider, location, waiting period and annual limit conditions always apply, and the linked insurer page and fund rules are the authority on all of them.

Some things are deliberately excluded. A rewards program may be listed as existing, but its points, caps and retail discount values are never quantified, never treated as a benefit amount, and never counted toward the member benefits note. Temporary promotions and sign-up offers are excluded entirely, as is any claim we could not verify on the insurer's own pages.

Many funds show no member benefits here. That can mean we have not yet researched or verified that fund, or that what it publishes did not meet the bar above, or that it publishes none. Larger funds tend to publish more of this material on the web, so absence says the least about small and restricted funds. As elsewhere on this page, silence means the published evidence does not support a statement, never that a fund is worse.

Where a product's expanded details list member benefits, results may also show a short note headed "Member benefits" stating, as fact, that published benefits exist for that fund, and pointing to the full list. What the note can show depends both on what a fund publishes and on how much of it we have verified, so read it as a statement about published evidence and not as a judgement of the cover. If you opt in to the member benefit questions, the note leads with the kind you said matters most. It never says a fund's benefits are strong or weak, and it never compares one fund's benefits with another's. The graded view is the benefits pills described below.

The benefits pills are our own assessment, and they are not a perfect science. We reviewed the services each insurer publishes, verified the evidence for every claim we count, and rated each kind of benefit for how much it typically delivers to members. An insurer's hospital and extras pills show how much of that rated ground their verified benefits cover. Only claims our evidence checks can currently vouch for are counted, so a pill can change as insurer pages change. Insurers offer many services beyond what we track; always confirm what matters to you with the insurer before acting.

Coverage here is uneven in a way worth knowing about. How many benefits we have found varies a lot from one insurer to the next, so a shorter list reflects how far our checks have reached at least as much as it reflects what a fund offers.

Insurers can tell us which page to check. If a benefit of yours is missing here, or the page we cite has moved, send us the page you would like our checks to point at. A link is enough, and no case needs to be made for it. The address is hello@clearchoicehealth.com.au

None of this changes any price shown, any market-value band, or the order of results. The value assessment described at the start of this page is computed from published premiums and published cover alone. Member benefit information appears beside that assessment, never inside it. Answering or declining the member benefit questions changes only how these facts are phrased and ordered, never which products appear or how they are assessed.

When we leave something blank explains what a blank means anywhere on this site, and what it does not mean about an insurer.

6. What this is not

• No product recommendations, no rankings, no "best" lists: grouped assessments only.
• No suitability or affordability claims: fit notes restate your own answers against published cover lists; they never predict what you will need, use or be paid.
• No statements about any insurer's profitability or conduct: a price above the market expectation is a fact about one product's price, nothing more.
• No sales relationship: no fund pays us, and the only outbound links go to the government's own product information statements.

7. When we leave something blank

A blank space here is almost always about us, not about the insurer. Usually it means we have not checked that fund yet, or that what we found was too thin or too old to be worth saying. We would rather leave a gap than tell you something we cannot stand behind.

So please do not read a blank as a black mark. Smaller funds tend to have more of them, simply because less is published about them and because we have not got to everyone yet.

There is one exception. The Government publishes a full list of every policy on sale, so when we say a policy does not cover something, we are reading that from the list. We never turn a gap in our own research into a fact about an insurer.

8. Data, freshness and provenance

All data comes from the Australian Government's PrivateHealth.gov.au dataset (published monthly; used under CC BY 3.0 AU). Results always state which release they were computed from. Until the model has seen a full annual market repricing, assessments are calibrated on a single market snapshot and carry a note saying so. They firm up as more market history accrues.

When we leave something blank explains what a blank means anywhere on this site, and what it does not mean about an insurer.

We use only the Australian Government's PrivateHealth.gov.au dataset for product data. Historical Medicare (MBS) statistics and ABS population estimates inform our population utilisation context, never product content or pricing. Member benefit notes are quoted from insurers' own public websites as dated, linked evidence, and never affect product data, pricing or value assessments. We do not currently use APRA or AIHW data, no commercial data of any kind, and nothing supplied to us by any insurer.

Your answers never leave your browser: the assessment runs on your device against the published data for your state.

9. Ombudsman complaint levels

ClearChoice compares each insurer’s share of complaints received by the Private Health Insurance Ombudsman with its share of Australian private health insurance policies. Lower is better. This is not a customer satisfaction survey.

Basis: Rolling 12 months of quarterly Ombudsman reports, covering October 2024 to September 2025. Market share as at 30 June 2024.

* Complaint results are reported jointly for this brand and other brands operated by the same registered insurer.

ClearChoice’s complaint-level indicator compares an insurer’s share of complaints received by the Commonwealth Ombudsman, acting as the Private Health Insurance Ombudsman, with the insurer’s share of Australian private health insurance policies. The indicator is independently calculated by ClearChoice Health.

It is not a direct customer satisfaction survey and does not measure every interaction with an insurer. Complaint levels may be influenced by fund size, unusual operational events, consumer awareness of the Ombudsman, the complexity of complaints, brand structure, internal complaint handling and changes to reporting or insurer ownership.

A complaint does not necessarily mean that the insurer acted incorrectly. The Ombudsman’s complaint categories include matters that may be finalised without further action, referred to an insurer or formally investigated.

ClearChoice uses the most recent official reporting period with sufficient validated data. When a fund is small, so are its complaint numbers, and we more often find we cannot make a fair comparison. The date range, source documents, confidence level and calculation method are published with each result. The Ombudsman’s complaint data is dynamic and may be revised in later publications.

The two directions are not held to the same test. A higher than expected complaint level is published only where the difference is unlikely to have arisen by chance, because a mark against a named insurer should rest on evidence. Where a difference does not meet that bar, the insurer is shown as broadly in line with its share of the market, and its figures are still published in the table above so you can see what the difference was. A lower than expected level is published either where that same test is met or where there is enough data to make a fair comparison.

ClearChoice’s categories are not ratings issued or endorsed by the Commonwealth Ombudsman, APRA, the Australian Government or any insurer.

Source: Commonwealth Ombudsman, Private Health Insurance Ombudsman quarterly updates and State of the Health Funds reports. Market share is reported from APRA Market Share, All Policies.

Calculated from these committed period files:

• quarter-2024-12.json
• quarter-2025-03.json
• quarter-2025-06.json
• quarter-2025-09.json

Official source documents:

• State of the Health Funds Report 2024-25
• Private Health Insurance Ombudsman Quarterly Update: June 2025
• Private Health Insurance Ombudsman Quarterly Update: September 2025

When we leave something blank explains what a blank means anywhere on this site, and what it does not mean about an insurer.

10. Digital app experience

Most people deal with their health fund through its phone app. ClearChoice relays the aggregate rating that app-store users have already published for each insurer’s Australian member app. ClearChoice does not rate these apps, does not test them, and does not survey members about them.

Where an insurer’s app is listed on more than one store, the figure is the average of the store ratings weighted by how many ratings each carries, so a store with ten times the ratings counts for ten times as much. The figure is shown to one decimal place.

How much a difference means depends on how many ratings sit behind it. A figure drawn from a few dozen ratings could sit most of a point either side of where it landed, so half a point between two small insurers may be nothing at all. A figure drawn from tens of thousands is settled to within about a hundredth, and there a tenth is real. The number of ratings behind each figure is shown alongside it, and it is the first thing to look at before comparing two of them.

These marks are shown only when enough of the insurers on your results page have a published figure. Showing them for a handful of insurers would say more about which listings ClearChoice has checked than about the apps themselves, so below that point no marks are shown at all.

App-store ratings are left voluntarily by people who chose to rate an app. They are not a random sample of an insurer’s members, and they are not a measure of the insurer itself: an app rating reflects the app.

Ratings are also shaped by things that have little to do with how well an app works, including how often an insurer prompts people to rate it, how long the app has been listed, and whether a developer has reset its ratings after a major release. The two stores compute their aggregates differently, so the figures are not exactly like for like.

ClearChoice records where each figure came from and when it was checked, and shows the figure only while it is recent enough to be worth comparing. Where an insurer publishes no member app, or its app has too few ratings, ClearChoice says so rather than filling the gap. Where ClearChoice has not identified or verified an insurer’s listing, nothing is shown for that insurer.

These figures are not issued, endorsed or reviewed by Apple, Google, the Australian Government, APRA or any insurer.

Source: aggregate customer ratings published on the Australian Apple App Store and Google Play listings for each insurer’s member app.

Method version 2026-08-a. Figures are recorded in the committed file src/data/app-ratings.json, which is reviewed before publication.

When we leave something blank explains what a blank means anywhere on this site, and what it does not mean about an insurer.