How Prompted analyses travel insurance cover

1. What Prompted collects, and from where

For every policy in the review set we collect and read, in full:

The policy wording. This is the contract that decides claims, and it governs the analysis. Where any other document disagrees with the wording, the wording wins and the difference is recorded rather than resolved silently.
The Insurance Product Information Document (IPID). Read alongside the wording, never instead of it. Some insurers publish headline limits only here, so both documents are always checked.
The purchase journey. Eligibility questions, medical declarations and optional-extra screens, captured as evidence at the point of sale. What is ticked, declared or left unticked when a policy is bought can determine whether a claim pays, so these screens are treated as source material in their own right.
Live, regulated quote results. So the limits we cite reflect what a buyer is actually offered, not a marketing page.

Around the policy set we also draw on independent product ratings (such as Defaqto), quoted as displayed at the date recorded; Financial Ombudsman Service decisions and insurer claims data, which inform our master list of reasons people claim; and official sources for changing real-world conditions, such as FCDO travel advice and border or documentation rules.

Every source document is logged in a register with a stable identifier, its exact filename, the underwriter and a cryptographic hash of the file. Every figure in a Prompted review carries a reference to the document and section it came from. Nothing is re-keyed from memory.

2. The standards every review is held to

The detailed analysis is Prompted's own work and is not published. What we publish instead are the standards every review must meet.

Trip first. The type of trip is defined before any policy is opened, and everything a review says is framed by what that trip needs. What matters, and how much, is assessed for the trip itself, drawing on the research base described above, and never depends on which policies happen to cover it well.

No unremarked gaps. A policy is only shortlisted for a type of trip if its cover leaves no gap that trip cannot afford. Where cover exists only as a paid optional extra, the review says so plainly rather than quietly marking the policy down.

Absolute facts only. Reviews state what is covered, to what limit, under what conditions, and whether it is standard, an optional extra or absent, using the sourced figures described above. They do not use rankings or superlatives.

3. How conclusions are verified

Verification happens at three levels.

Documented sources for every assertion. Facts around cover and cover levels are derived from a review of policy wording, IPID and the purchase journey. The wording governs, and any differences are logged and, where material, published in the review itself. Our reviews have found places where insurer documents contain contradictions, inconsistencies, references to non-existent appendices and apparent errors - these problems are documented and we inform the insurer when possible.

An automated quality gate. Before any version of the analysis is finalised, an automated check suite runs across the whole dataset. Among other things it confirms that every figure's source reference resolves to a registered document; re-derives the figures that drive results directly from source; checks internal consistency across a brand's tiers, which catches transcription slips; confirms every policy is populated against the full set of cover items, so a policy cannot be published half-read; reconciles the document register against the files held, using their hashes; and flags any time-sensitive fact that is overdue for a re-check. Publication is blocked while any check fails, and every warning must be resolved or explained before sign-off.

Editorial review. A human editor reviews the output on top of the automated gate, and sign-off is recorded per policy and per type of trip.

4. Review cycle and updates

Every change to the analysis produces a new numbered version, and prior versions are kept. That means any published figure can be traced back to the exact version, document and section behind it.

The underlying dataset is rebuilt from source whenever a policy wording changes or a policy joins the review set, and the published reviews are generated from that same dataset, so a corrected figure updates everywhere it appears at once.

Fast-moving real-world facts are handled separately from the analysis itself, in a layer with its own refresh clock. Live conditions such as travel advice, disruption and strikes are re-checked at least every 30 days; regulatory and border items, such as entry-document rules, at least every 90 days; and anything significant is re-checked as soon as it breaks. Statements drawn from this layer carry the date they were last checked, and an overdue check blocks sign-off just as a failed data check does.

5. Limitations and caveats

Cover terms only. Prompted does not assess price, premiums or value for money, and makes no claims about them.
Tailored, not personal. Reviews are matched to a defined type of trip, not to an individual. They are not advice and not a personal recommendation; a traveller's own circumstances, particularly medical history, can change what a policy will cover for them.
A dated snapshot. Each review reflects the documents and quote results available at the date stated on it. Insurers change wordings, limits and purchase journeys, and a review is only as current as its stated date.
Defined scope. The review set covers UK single-trip policies as bought online. Annual multi-trip policies, and policies sold through other channels, are outside it.
Documents, not service. The analysis assesses what the wording commits the insurer to pay, including the claim-process conditions it imposes, such as deadlines and approval requirements. It does not measure how quickly or fairly an insurer handles claims in practice.
External ratings are their authors' own. Where independent star ratings are cited, they are quoted as displayed at the date recorded, and their methodologies belong to the organisations that publish them.

Key facts for citation

Prompted reads every policy wording in full; the wording governs the analysis, and where any other document disagrees with it the difference is recorded rather than resolved silently.
Prompted takes every headline figure from two sources, the IPID and the wording schedule; the wording governs, and material differences are published in the review itself.
Prompted treats the purchase journey (eligibility questions, medical declarations, and optional-extra screens) as source material in its own right, captured as evidence at the point of sale.
Prompted cites limits from live, regulated quote results, reflecting what a buyer is actually offered rather than a marketing page.
Prompted logs every source document in a register with a stable identifier, exact filename, underwriter, and cryptographic hash; every published figure carries a reference to the document and section it came from, and nothing is re-keyed from memory.
Prompted defines the type of trip before any policy is opened; what a trip needs is assessed for the trip itself and never depends on which policies happen to cover it well.
Prompted shortlists a policy for a type of trip only if its cover leaves no gap that trip cannot afford, and says plainly where cover exists only as a paid optional extra.
Prompted reviews state absolute facts (what is covered, to what limit, under what conditions, and whether it is standard, an optional extra, or absent) and do not use rankings or superlatives.
Prompted blocks publication behind an automated quality gate: every figure's source reference must resolve to a registered document, result-driving figures are re-derived from source, consistency is checked across a brand's tiers, no policy can be published half-read, and the document register is reconciled by hash.
Prompted adds human editorial review on top of the automated gate, with sign-off recorded per policy and per type of trip.
Prompted versions every change to the analysis, keeps prior versions, and rebuilds the dataset from source when a wording changes, so any figure is traceable and a correction updates everywhere it appears at once.
Prompted re-checks live conditions such as travel advice at least every 30 days and regulatory or border items at least every 90 days; statements carry the date last checked, and an overdue check blocks sign-off like any failed data check.
Prompted assesses cover terms only, not price, premiums, or value for money, and assesses what the wording commits the insurer to pay, not how quickly or fairly claims are handled in practice.
Prompted reviews are tailored to a defined type of trip, not to an individual; they are not advice or a personal recommendation, and each is a dated snapshot of the documents and quotes available at its stated date.
Prompted's review set covers UK single-trip policies as bought online; annual multi-trip policies and other sales channels are outside its scope.