Methodology
This page documents how Phidea scores the tools it covers. It is binding. Every score on the site can be reproduced by applying the rules here to the sources linked on the tool’s fiche.
What the score is
A tool’s score is a ratio. Four axes, each scored 0 to 5, each weighted equally. Axes for which Phidea cannot find a public source are excluded from both the numerator and the denominator; a tool with three sourced axes scores out of 15, not 20.
The score is a summary of observable, verifiable signals at a point in time. It is not a recommendation, and the editorial copy on this site is careful to treat it as evidence rather than as verdict.
The four axes
1. Traction — named carrier deployments
The count of named carrier customers whose adoption of the tool can be verified against a public primary source. A primary source is a press release from the carrier or the vendor, a case study on either party’s website, an independent trade-press article naming the carrier, or a disclosure in a regulatory or SEC filing.
| Named carriers | Score |
|---|---|
| 0 | 0 |
| 1-2 | 1 |
| 3-5 | 2 |
| 6-10 | 3 |
| 11-20 | 4 |
| 20+ | 5 |
What this captures. A carrier that publishes a name speaks with its reputation. The axis is a proxy for enterprise willingness to defend the vendor in writing.
What this does not capture.Total market share. Large carriers often use tools they never name publicly. A “trusted by 200+ insurers” banner without names does not contribute to this axis; an investor-customer relationship disclosed in a Series D press release does.
Aggregate claims.Where a vendor publishes an aggregate only (“22 of the top 25 US property insurers use our product”), the axis reflects the count of individually named carriers Phidea can separately verify. This depresses the score of incumbents like Verisk Xactware whose market position is real but whose per-customer case studies are thin. The distortion is documented, not corrected for.
2. Maturity — years since founding
Calendar years between the vendor’s founding year, as established in Crunchbase, CB Insights, or another primary source, and the current year.
| Years since founding | Score |
|---|---|
| < 2 | 0 |
| 2 – 4 | 1 |
| 5 – 6 | 2 |
| 7 – 9 | 3 |
| 10 – 14 | 4 |
| 15+ | 5 |
What this captures. Age is a proxy for the breadth of edge cases the product has handled and the depth of its integration surface with adjacent enterprise systems.
What this does not capture. Product quality. Nor whether the vendor is still shipping. A 25-year-old product that has not had a meaningful release in a decade scores 5 on this axis; the aggregate contextualises this because it is one of four.
Rebrands and acquisitions.Maturity follows the product’s architectural lineage, not the legal entity’s incorporation date. CCC Intelligent Solutions is scored at 1980 (product origin), not 2021 (SPAC close). Xactware is scored at 1986, not 2006 (Verisk acquisition).
3. Coverage — insurance lines supported
The count of distinct insurance lines the product supports, restricted to lines where the vendor has at least one named customer or a public feature announcement specific to that line.
Recognised lines: auto, home, commercial, life, health, workers’ compensation, specialty, reinsurance.
| Lines supported | Score |
|---|---|
| 0 | 0 |
| 1 | 1 |
| 2 | 2 |
| 3 | 3 |
| 4 – 5 | 4 |
| 6 or more | 5 |
What this captures. Breadth is a proxy for productisation. A product that works across six lines has usually abstracted away line-specific logic, a capital-intensive thing to do.
What this does not capture. Depth. A vendor claiming six lines with one customer each scores higher than a vendor with ten customers in one line.
4. Analyst recognition — coverage in primary analyst houses
Citations from firms whose editorial output is treated as authoritative by the enterprise buyers Phidea writes for.
- Insurance-specific: Celent, Gartner, Forrester, Novarica, Insurance Thought Leadership.
- Adjacent (used for IDP, BPO, and claims-workflow categories): Everest Group, ISG, Deep Analysis, HFS Research.
A citation is classified major when the vendor is positioned inside a flagship recurring product — Gartner Magic Quadrant, Forrester Wave, Everest PEAK Matrix, ISG Buyers Guide, Celent Vendor Matrix — with an assigned position (Leader, Strong Performer, Challenger, Contender, or an equivalent). Anything else is classified minor: analyst notes, vendor profiles, panel mentions, conference appearances.
| Citations | Score |
|---|---|
| None | 0 |
| 1 minor | 1 |
| 1 major, or 2 minors | 2 |
| 1 major + 1 additional | 3 |
| 2 majors | 4 |
| 3 or more majors | 5 |
What this captures. Analyst houses are the gatekeepers of enterprise procurement. A vendor that survives a Magic Quadrant has opened its books to a credentialed third party.
What this does not capture. Product quality on its own. The axis is a strong signal for enterprise readiness and a weak signal for technical superiority.
Why equal weights
The four axes capture distinct things. Weighting one above another would embed an editorial opinion inside the score. Readers who care more about one axis than another should look at the axis-level breakdown on the tool page rather than compare aggregates.
Ranking eligibility
A tool with fewer than three sourced axes is listed as a profile and flagged “insufficient signals”. It does not appear in the ranked order of comparison pages. The threshold is a hard floor, not a gradient: a two-axis score is noisy enough to mislead.
Inclusion policy
Any vendor serving regulated industries may be added to the database. Phidea does not charge for inclusion. There is no paid placement, no advertising, no sponsored listings, no “enhanced” tiers. If any of this changes, the page carrying the change will be marked explicitly and affected fiches will carry a disclosure banner.
Vendors may request inclusion by emailing benoit@phidea.fr with a case study or press release. The methodology applies equally to requested additions and to additions Phidea initiates.
Update cadence
Every tool page carries a “last verified” date. Each fiche is intended to be re-verified at least once per quarter, and is re-verified immediately when a source link rots, a vendor notifies Phidea of a material change, or a news item on this site mentions the tool. When re-verification changes an axis score, the fiche updates silently and the aggregate shifts.
Methodology versioning
The current version is 1 (published 2026-04-21). Any change to the axes, bands, or aggregation formula triggers a numbered revision. When a revision lands, existing fiches are re-scored under the new methodology; previous aggregate scores remain visible in the fiche’s history. Revision notes describe what changed and why, linked from this page.
What the methodology does not do
The methodology measures what outsiders can verify. Four things sit outside that boundary, and the absence of each shapes how a score should be read.
- Product quality. A well-scored product can be a poor choice for a specific carrier, and an under-scored product can be the right one. The signals Phidea aggregates are visible to outsiders; product quality is not.
- Pricing. Vendor pricing in enterprise insurance is rarely public, routinely negotiated per carrier, and too noisy to compare across customers.
- User sentiment. Anonymous reviews on G2/Capterra-style platforms are not a reliable signal for enterprise-insurance buyers. The methodology does not ingest them.
- Proprietary analyst insight. Where an analyst’s strongest assessment sits behind a paywall, Phidea can only see the public edge — the press release, the vendor-hosted excerpt, the public position statement.
Corrections
A misattributed deployment, an outdated funding round, a source link that rots, a rebrand missed — submit a correction. Two channels are open:
- Open a pull request against github.com/benoitp4squier-lang/phidea-tracker. The correction lives alongside the source and gets merged in public.
- Email benoit@phidea.fr with the source URL and the field that needs to change.
Both channels trigger a redeploy; corrections land on the site within minutes of the merge.