Methodology
How we measure
We measure observable messaging on vendor homepages and official press releases. Not security quality, product effectiveness or trustworthiness.
Definitions
- Cohort
- Explicit vendor list and active status for each period. Use a stable denominator or disclose cohort changes.
- Homepage adoption of term T in period P
- Number of eligible vendors with T detected in the designated latest successful homepage snapshot within P / number of eligible vendors with valid homepage snapshots for P. Display numerator and denominator; if the intention is to report “43 of 50,” all 50 must have valid evidence for the reporting period or missing coverage must be explicitly shown.
- Press-release vendor adoption
- Number of vendors with at least one eligible official release containing T in period P / number of vendors with confirmed release monitoring coverage in P. Zero releases is not automatically zero usage if coverage is unknown.
- Occurrence count
- Matches after alias normalization and deduplication within the defined text field/corpus; disclose whether headlines, body and boilerplate are included.
- Density
- Occurrences per 1,000 words in a defined source and window. Homepage denominator = extracted homepage text length; press releases use an explicitly defined release corpus and deduplication rule. Do not compare unlike corpora without labeling.
- Velocity
- Difference in adoption percentage points between comparable periods; avoid labeling this percentage growth. Show sample sizes.
- First observed
- Earliest successful captured occurrence in available data; never claim first-ever public use.
- Hero-change count
- Count verified distinct meaningful headline versions first observed in a period, with capture-gap limitations.
- Tracked vocabulary overlap
- Jaccard intersection/union of detected dictionary terms for a defined period/source. This is a narrow vocabulary measure, not proof of message interchangeability.
- Co-occurrence
- Two terms within the same defined text unit (e.g., press-release body or homepage hero), not simply anywhere in a vendor's lifetime corpus.
- Reporting calendar
- Choose UTC periods and a consistent monthly cutoff. Define how snapshots near month boundaries are handled.
Limitations
- Failed or missing captures are shown as missing, never as zero.
- Homepage and press-release data are separate datasets with separate denominators.
- Hero claims are selected by our editorial team as vendors might have multiple at the moment of evaluation.