Skip to main content

How Kave selects and ranks resources

All of it, including the numbers. If Kave will not show you how a ranking was produced, the ranking is not worth trusting.

Nobody can pay to be here

Kave accepts no payment, commission, referral fee, or affiliate revenue from any organisation it lists, and there is no field in the database that could record such a relationship. This is a structural commitment rather than a policy one: the ranking functions have no input that could encode payment, which you can verify by reading supabase/migrations/0003_search.sql in the source.

Four gates that come before any score

A resource must pass all four. Failing any one excludes it no matter how well it would otherwise score — which is the point, because a high-impact program that hides its limitations is exactly the kind of listing a score alone would let through.

  1. Meaningful and direct patient value
  2. Verifiable access, eligibility, or pricing
  3. Clearly disclosed limitations
  4. No paid influence over inclusion or ranking
Quality score criteria and weights
CriterionWeightWhat it measures
Patient impact25How much this changes a real outcome: care received, a bill reduced, a prescription filled.
Affordability or financial value20Size and reliability of the financial benefit to the user.
Accessibility15Whether people can actually reach it: language, disability, connectivity, documentation burden.
Transparency15Whether eligibility, cost, and limitations are published before someone commits.
Ease of use10Steps, wait time, and how hard the application is to finish.
Geographic reach5Deliberately small. A strong single-state resource should not lose to a weak national one.
Institutional credibility5Track record and accountability of the operating body.
Privacy and data practices5What the program collects, shares, and sells.
Inclusion threshold80Out of 100, in addition to passing all four gates.
  • Why geographic reach is worth only five points

    A directory that rewards national reach ends up recommending weak national programs over strong local ones. A free clinic serving one county can be the best answer available to the person asking, so reach is deliberately the second-smallest term in the score.

  • How the matching engine ranks your options

    Matching is deterministic. There is no model, no personalisation, and no randomness: the same answers always produce the same three results, in the same order. First, hard filters remove anything that cannot serve you — wrong category, wrong state, wrong age, an insurance status the program cannot accept, or a delivery mode you ruled out. Answering “I do not know” or skipping a question never filters anything out.

    Whatever survives is scored out of 100:

    • Category fit — 30
    • Geographic fit — 20
    • Insurance fit — 15
    • Specific need fit — 15
    • Timing and availability — 10
    • Delivery preference — 5
    • Kave quality score — 5

    Then a diversity rule applies: the three results must be three different kinds of solution. Three manufacturer copay cards are one answer wearing three hats, so Kave shows the strongest direct solution, an alternative with different eligibility or access, and where one exists a local or safety-net backup. Everything else that matched is listed below them — nothing that fits is hidden.

  • How often things are re-checked
    • Automated link check — every 7 days
    • High-impact program review — every 30 days
    • Standard substantive review — every 90 days
    • Immediately, when a material change is detected

    Staleness is computed rather than stored. If a review window closes or an official link stops resolving, the badge changes on every page at once, without waiting for a background job to catch up. A record cannot keep displaying “Verified” because a job failed.

  • What Kave refuses to do
    • Publish an eligibility threshold that no person has confirmed against an official document. Automated extraction may draft a field; it may never publish one.
    • Rate or rank hospitals by quality. The CMS import deliberately discards the star ratings in the source file.
    • Show star ratings, review scores, or testimonials for any program.
    • Ask for a diagnosis, a Social Security number, a date of birth, an insurance member ID, or a hospital account number.
    • Store your answers to the guided questions.

What the verification labels mean

Verified
A Kave reviewer checked this hospital's official policy, application, and contact information within the current review period.
Partially verified
A Kave reviewer confirmed some official documents, but one or more important fields are still incomplete. Confirm the details with the hospital.
Needs re-review
The policy changed, a link broke, or the review period expired. Treat the details below as out of date until Kave re-checks them. Structured eligibility figures are hidden in this state — a stale threshold is worse than none, because it stops people applying who would have qualified.
Not yet verified
This hospital is searchable, but Kave has not completed a substantive review of its financial-assistance policy. Kave will not guess at eligibility rules. Structured eligibility figures are hidden in this state — a stale threshold is worse than none, because it stops people applying who would have qualified.

Found something wrong?

Reports from users are one of the main ways records get corrected. Report outdated information — it takes about thirty seconds and no account.