The Insurf True-Cost Score.
Specification v1.0 · published August 28, 2026
One number for what a health plan is simulated to cost a specific person over twelve months — defined publicly, computed deterministically from state-filed data, and versioned so anyone can check it. This page is the specification.
The definition.
The True-Cost Score for a person × plan is the simulated 12-month total cost, in dollars: the plan's premium — net of any premium tax credit when income is provided — plus expected cost-sharing (deductibles, copays and coinsurance including drug tiers, capped by the plan's out-of-pocket maximum) under the person's stated care profile.
It is a dollar figure, not an opaque index — deliberately, so it can be checked.
Premium
Twelve months of the plan's filed premium for the person's age, county rating area, and tobacco status where states rate that way — net of any premium tax credit when income is provided.
Expected cost-sharing
The person's stated care profile replayed through the plan's filed accumulators: deductibles (medical and drug channels kept separate, as filed), copays and coinsurance, including drug tiers.
The cap
Cost-sharing is capped by the plan's filed out-of-pocket maximum, so a Score can never exceed twelve months of premium plus the plan's own filed cap.
The output
One dollar figure per person × plan. Not a grade, not a percentile, not a proprietary scale — arithmetic that can be checked against the filings it came from.
The computation discipline.
The Score inherits the cost engine's rules in full — restated here because a score is only as trustworthy as what it refuses to guess.
State-filed public data
Every Score is computed from state-filed public plan data — 15,062 filed plan variants across sixteen states in the live artifact. The full source manifest, named filing by filing and checksum-verified, is published on the pricing methodology page.
Snapshot-pinned
Computations run against pinned data artifacts with recorded snapshot IDs, so any published figure traces to the exact bytes that produced it. A download timestamp is never presented as the data's vintage.
Fail-closed
Copays are never guessed. Ambiguous filed cost sharing — a source string that could be a copay or coinsurance — stays flagged as conflicting instead of being resolved by assumption. A covered drug's dollar estimate is withheld until its exact plan-level tier placement is known. These limits never become silent zeros — they lower precision visibly or withhold a number entirely.
One labeled fallback rule
Where benefit-level copays are not applicable in the filed data, a uniform 25% post-deductible coinsurance rule is applied — always labeled and disclosed where used — so plan differences come only from filed premiums, deductibles, and out-of-pocket maxima, never from invented copays.
Deterministic
Identical inputs against the same snapshot produce identical output. Determinism is what makes the version number on this page meaningful — and what makes a published figure something you can replay rather than something you have to trust.
The full source manifest and checksum-verification procedure live on the pricing methodology page.
What the Score is not.
Not a prediction of health events
The care profile is an input, stated by the person. The Score prices that stated year against each plan's filed terms; it does not forecast what will actually happen to anyone's health.
Not a plan recommendation by itself
The Score orders plans by simulated 12-month cost under the captured assumptions. Doctors, drugs, networks, and eligibility still matter — a licensed human and the member make the plan decision.
Not a quality or network-adequacy rating
A low Score says nothing about a carrier's service, its provider directory's accuracy, or the adequacy of its network. The Score measures modeled cost — only that.
Every figure computed with the Score is a modeled illustration under labeled assumptions — never a quote, and never a statement of what any individual will actually spend.
Check the work.
Every published Score figure derives from a committed evidence artifact — the deterministic output of a published method over a pinned snapshot. The figures on this page are computed from those artifacts at build time, never hand-typed.
Worked example · plan-choice statistics
The cost of premium-sorting, measured with the Score across all 159 Georgia counties — 4,293 county-profile-age combinations of 2026 filed plans, 0 skipped, snapshot snapshot_8827361338bd…, source as of 2026-06-03.
Worked example · the Georgia Renewal Shock Index
The Score applied county by county across Georgia: the cheapest filed 2026 plan at age 40 and the median annual cost of premium-only shopping in each county, from the same committed artifact family.
Find an error in a published figure and we will correct it and log it — corrections are recorded in the changelog below, alongside version history.
Versioned. No silent revisions.
A methodology change always produces a new version entry here. A figure published under v1.0 stays attributable to v1.0 — the definition never shifts underneath a number that cited it.
v1.0
August 28, 2026
Initial publication: the definition, the fail-closed computation discipline, the labeled 25% post-deductible coinsurance fallback, and the verification trail — as shipped in the live engine.
Questions the spec should answer.
What is the Insurf True-Cost Score?
The simulated 12-month total cost of a specific health plan for a specific person, in dollars: the plan's premium (net of any premium tax credit when income is provided) plus expected cost-sharing — deductibles, copays and coinsurance including drug tiers — under the person's stated care profile, capped by the plan's out-of-pocket maximum. It is deliberately a dollar figure rather than an opaque index, so every Score can be checked against the filed data that produced it.
How is the Score different from sorting plans by premium?
Premium is one component of a year of costs — and the only one most shopping tools rank by. The Score prices the whole year. Measured with the Score across Georgia's 2026 filed plans (every county, nine illustrative care profiles, three ages — methodology and evidence artifact published), the lowest-premium plan was the wrong pick 57% of the time, costing a median $548 a year versus the lowest-Score plan.
Can anyone replay a published Score figure?
That is the point of the discipline: the computation is deterministic over snapshot-pinned, state-filed public data, so identical inputs against the same snapshot produce identical output, and every published figure derives from a committed evidence artifact that records its snapshot ID and method. The worked examples — the plan-choice statistics and the Georgia Renewal Shock Index — publish their scope and assumptions. Find an error in a published figure and we will correct it and log it.
Does compensation affect the Score?
No. The Score is arithmetic over filed public data; compensation never enters the computation, and the same inputs produce the same Score regardless of which plan or carrier ranks first. When enrollment happens through the individually licensed insurance producer who supports Insurf users, carriers may pay that producer a commission, and that compensation is disclosed on estimates — the Score's math never varies with compensation.
The spec is public so the number can be checked.
Insurf, Inc. is not a licensed insurance agency; enrollment support is provided by an individually licensed insurance producer, who may be compensated by carrier commissions when an enrollment occurs — that compensation is disclosed on estimates, and the Score's math never varies with compensation. Insurf is not affiliated with or endorsed by any carrier, Georgia Access, or any state. Figures on this page are modeled illustrations from filed public data under the labeled assumptions above — informational only, not an offer of coverage, not a recommendation to buy or drop any plan, and not financial, tax, legal, medical, coverage, or insurance advice.