Skip to main content
Methodology

Every number, and where it came from.

Koala sells no insurance and takes no commission, so it has nothing to gain from the number it shows you. This page is the whole model: the published datasets it prices from, the exact arithmetic applied to them, and the things it cannot see. Read it and you can check our work.

See a worked estimate Jump to the sources

Read live from the rate provider the engine computes with, so this strip cannot advertise a dataset it no longer uses.

What the model covers

Lines of insurance priced
5Lines of insurance pricedAuto, homeowners, renters, health, and pet: each with its own dataset
Published datasets
4Published datasetsNAIC, III, KFF, and NAPHIA: all public, all citable, none from a carrier
States with seeded rate data
42States with seeded rate dataEverywhere else falls back to the national average, and says so
Coverage levels modeled
12Coverage levels modeledEach with its price factor and coverage basis disclosed in full
Price drivers Koala does NOT model
7Price drivers Koala does NOT modelCounted, named, and published, because an estimate is only as honest as its gaps
Why these numbers

You can check every number yourself.

Koala is new. It has no track record to point at, so it does not ask you to take anything on faith. It shows the working instead.

The data is public

Every base figure comes from a regulator or an industry body that publishes it openly: NAIC, the Insurance Information Institute, KFF, NAPHIA. Not a carrier feed, not a rate card someone gave us on terms. You can open the source and find the number.

The arithmetic is published

Koala shows the base figure, the coverage factor, the state adjustment, and the band, all as arithmetic on the estimate itself. There is no second model behind the one you can see, and no step we do not print.

Nobody pays us for the answer

Koala earns nothing from which policy you pick. It sells no insurance, receives no commission, and has no carrier relationships to protect. The incentive that bends every other comparison site does not exist here.

Read this partKoala's rate tables are seeded approximations today, not a live feed.

The figures below approximate each publisher's reported averages closely enough to be useful for comparison, and they are order-of-magnitude accurate, but they are a seed dataset, and Koala says so on every estimate rather than implying a precision it does not have. The engine reads them through a swappable interface, so connecting a live rate feed replaces the numbers without changing a line of the arithmetic on this page.

The datasets

One published source per line.

Each block below is read straight from the dataset the estimate engine actually loads. If a source, a year, a band, or a coverage factor changes, this page changes with it. Nothing here is hand-copied.

NAIC / IIIAutoas of 2023

NAIC & III state average annual auto insurance expenditure

Open the source

Base figure

$1,700/yr

The national average, used when a state has no seeded figure

States seeded

30

Everywhere else falls back to the national figure

Estimate band

±20%

Around the midpoint, per vehicle

Coverage levels

  • State minimum · ×0.42
  • Standard · ×1
  • Full coverage · ×1.32

Seed approximation of published state averages; replaced by the live rate feed once connected. State minimum and full-coverage levels are derived from the standard-coverage average using the coverage factors shown.

What Auto coverage actually is
IIIHomeownersas of 2023

Insurance Information Institute state average annual homeowners (HO-3) premium

Open the source

Base figure

$1,900/yr

The national average, used when a state has no seeded figure

States seeded

29

Everywhere else falls back to the national figure

Estimate band

±22%

Around the midpoint, per policy

Coverage levels

  • HO-3 (standard) · ×1
  • HO-5 (broader) · ×1.22

Seed approximation of published state HO-3 averages; replaced by the live rate feed once connected. Coverage-to-price uses the dwelling's insured value (Coverage A) shown in the assumed factors.

What Homeowners coverage actually is
IIIRentersas of 2023

Insurance Information Institute state average annual renters premium

Open the source

Base figure

$180/yr

The national average, used when a state has no seeded figure

States seeded

25

Everywhere else falls back to the national figure

Estimate band

±20%

Around the midpoint, per policy

Coverage levels

  • Standard · ×1
  • Extended · ×1.45

Seed approximation of published state renters averages; replaced by the live rate feed once connected.

What Renters coverage actually is
KFFHealth (ACA)as of 2024

KFF Health Insurance Marketplace benchmark (2nd-lowest silver) premium

Open the source

Base figure

$5,724/yr

The national benchmark for a 40-year-old, before premium tax credits

States seeded

23

Everywhere else falls back to the national figure

Estimate band

±15%

Around the midpoint, per person, age-rated from 40

Coverage levels

  • Bronze · ×0.8
  • Silver · ×1
  • Gold · ×1.15

Seed approximation of published marketplace benchmark premiums for a 40-year-old, before premium tax credits; replaced by the live rate feed once connected. Metal tiers and per-member ages derive from the benchmark using the factors shown.

What Health (ACA) coverage actually is
NAPHIAPetas of 2023

NAPHIA State of the Industry national average annual pet premium

Open the source

Base figure

Per species

NAPHIA publishes a rate per species per plan, so there is no base figure to scale

States seeded

National only

This publisher reports nationally, so this line is not state-adjusted

Estimate band

±25%

Around the midpoint, per pet

Coverage levels

  • Accident only
  • Accident & illness

Seed approximation of published national averages by species and plan; NAPHIA reports nationally, so pet estimates are not state-adjusted. Replaced by the live rate feed once connected.

What Pet coverage actually is
Worked end to end

One estimate, with nothing left out.

Below is a real estimate for a sample household (two cars, two pets, a home in Texas), computed live by the same engine that runs on your profile. Every step it took is printed.

Auto: Standard

Sample household · Texas · a 2019 Toyota RAV4 and a 2015 Honda Civic

$2,886 to $4,330

per year · ≈$301/mo at the midpoint

What this assumed

State
state-level average for TX
Vehicles
2 (multi-vehicle factor ×1.85)
Coverage level
Standard

The arithmetic

  1. 1Auto standard base (state-level average for TX): $1,950/yr per vehicle
  2. 2× coverage factor for Standard: ×1
  3. 3× multi-vehicle factor (2 vehicles): ×1.85
  4. 4= estimated annual mid-point $3,608, shown as a ±20% range

The range shown is the midpoint ±20%. That band is not a confidence interval; it is the spread published rate data implies, and a real quote can land outside it.

Coverage-to-price metric

coverage-to-price = $500,000 (assumed 100/300/100 liability + comprehensive/collision, total limits $) ÷ $3,608 (estimated annual price) = 138.6

Higher means more assumed coverage per dollar. Comparable only within this line: a computed metric, never a recommendation.

Source: NAIC & III state average annual auto insurance expenditure · data as of 2023

An estimate, never a quote

That is the whole calculation. There is no second model behind it. Build your profile to run it on your own household, or see the worked sample estimate.

The limits

What Koala cannot see.

An estimate is only as honest as the list of things it ignores. These are the price drivers Koala's model does not include. Every one of them can move your real premium, and none of them is in the number we show you.

Auto

  • Your driving record, credit, and claims history

    These move a real quote enormously, and Koala does not model them. That is exactly why what you get is a range from published averages and not a quote.

Home

  • Roof age, claims history, and your deductible

    Real underwriting leans hard on all three. Koala models none of them, which is why the output is a band, not a price.

Renters

  • Replacement cost versus actual cash value

    Whether a stolen four-year-old laptop pays out as a new laptop or a four-year-old one. Koala's tiers assume replacement-cost thinking; a real policy may not, and it's the first thing to check.

Health

  • Your income, and therefore your subsidy

    This is the big one, and Koala does not model it. Premium tax credits are calculated against the benchmark silver plan and your household income, and they can reduce what you actually pay to a fraction of the sticker price, or to nothing. Treat every health number Koala shows as a pre-subsidy figure.

  • Tobacco use and household composition

    Both move real marketplace pricing. Neither is modeled here.

Pet

  • Breed and age at enrollment

    Both move real pricing sharply: some breeds are underwritten far more expensively, and premiums climb as a pet ages. Koala models neither.

  • Your reimbursement rate, deductible, and annual limit

    The three dials that decide what you actually get back. A cheap policy is often a 70% reimbursement with a low annual cap, and that is a different product from an expensive one.

Estimated range based on published state/national rate data for similar profiles. This is not a quote and not an offer of insurance. Actual pricing depends on the specific insurer's underwriting and may differ from this estimate.

Keep reading

Now run it on your own household.

You have seen the sources, the arithmetic, and the gaps. Build your profile once and Koala estimates every line you carry, showing its working every time.

Build your profile See a sample estimate
Free mode: no payments, nothing is charged