The B2B SaaS AI Visibility Benchmark 2026

Which B2B SaaS brands do AI answer engines name when a buyer asks for software? 94 brand entries (93 companies) across five G2 categories, 50 buying questions frozen before collection, five AI platforms (ChatGPT, Claude, Perplexity, Google Gemini, Google AI Overviews), September 2026.

Headline: 32 of 94 brands (34%) were named on none of the five platforms. Of 917 domains consulted by ChatGPT, Claude and Perplexity on the same 50 questions, 71.4% were consulted by exactly one engine, most of them consulted just once.

Version 1.5 recounts every consulted, retrieved and cited figure from the raw captures. Three headline figures moved: the one-engine share (71.6% to 71.4%), the vendor share of citations (30.7% to 34.4%) and the retrieved-to-cited rate (43.5% to 40.7%). The 32 of 94 headline did not. Corrected: all 50 ChatGPT prompts carried the year and the other engines' prompts did not, which earlier editions misdescribed; two ChatGPT chats now count their final attempt only; cited now means cited in the answer text, with image carousels and thumbnails published apart; six vendor domains moved from third party to own; text in the Claude and Perplexity transcripts that was not a source removed from the domain counts (940 to 917); cited-domains uncapped; adjudication before and after columns restored. New: the wording ChatGPT received, source domains per question, the ChatGPT funnel per brand and question, each vendor's own domains, the tracker source rows the report quotes, SHA-256 fingerprints, and verify_benchmark.py, which recomputes the main numbers from these files and lists what it cannot check.

Full report and method: https://citant.ai/research/b2b-saas-ai-visibility-benchmark-2026/