# AI Visibility Index — Data Dictionary

One line per column in `ai-visibility-index-dataset.csv` (68 vendors). Collected August 2026. Licence: CC BY 4.0.

## Identity
- **vendor** — brand / product name.
- **category** — study category: `helpdesk`, `sales_engagement`, or `product_analytics`.
- **role** — functional role tag (e.g. engagement, data, intelligence, helpdesk, ecommerce, product_analytics).
- **tier** — `incumbent` or `challenger`.

## AI visibility (the outcome)
- **visibility_openai / visibility_claude / visibility_gemini** — % of that model's 39 category responses (13 prompts × 3 repeats) in which the vendor was recommended.
- **visibility_pooled** — % across all 117 category responses (the 3 models combined).

## Provenance
- **official_domain / homepage** — the vendor's primary site.
- **founded_year** — year founded (where known).
- **g2_url / capterra_url / linkedin_url** — profile URLs used for collection.
- **collected_on** — date the signal metrics were gathered.

## SEO / Layer-1 signals (Ahrefs, US)
- **ahrefs_dr** — Ahrefs Domain Rating (0–100).
- **organic_traffic** — estimated monthly organic visits.
- **referring_domains** — number of linking domains.
- **branded_search_volume** — US monthly search volume for the brand name.

## Third-party / Layer-2 trust signals
- **g2_reviews / capterra_reviews / trustradius_reviews** — review counts on each platform.
- **g2_rating / capterra_rating** — average star rating (out of 5) on each platform.
- **total_thirdparty_reviews** — sum of the three review counts.
- **avg_thirdparty_rating** — mean of G2 and Capterra ratings.
- **linkedin_followers / twitter_followers / youtube_subscribers** — audience counts.
- **youtube_video_count** — number of videos on the brand's YouTube channel.
- **reddit_mentions** — approximate `site:reddit.com "brand"` result count (rough proxy).
- **wikipedia_exists** — `yes`/`no`: does the brand have a Wikipedia page.
- **social_following_index** — composite: mean of the z-scored log-follower counts for LinkedIn + X (equal weight, so the larger platform doesn't dominate). This is the value used in the analysis; the individual follower columns are kept above for transparency.

## Listicle presence (third-party "best [category]" roundups)
Measured by scanning the top organic results of five "best [category]" query variations per engine, keeping independent editorial roundups, and scoring how many named each vendor.
- **google_listicle_appearances / bing_listicle_appearances** — # distinct qualifying listicles naming the vendor, per engine.
- **google_listicles_scanned / bing_listicles_scanned** — # qualifying listicles scanned in the vendor's category, per engine (the denominator; differs by category/engine).
- **google_listicle_coverage_pct / bing_listicle_coverage_pct** — appearances ÷ scanned, per engine (0–1).
- **listicle_appearances_total / listicles_scanned_total / listicle_coverage_pct_combined** — the same, pooled across both engines.
- **listicle_collected_on** — date listicle data was gathered.

## Notes
- Blank cells mean the metric could not be confidently identified for that vendor (no matching profile/account), not zero.
- Live web/social/SERP figures drift over time; treat these as an August 2026 snapshot.
- Full method: see the study page and `listicle_methodology.md`.