Skip to content

DABYTE DATA DESK

How the DABYTE AI Visibility Index is measured

Answer. A fixed panel of 16 category buyer prompts is submitted to 3 AI engines (openai, perplexity, gemini). For each answer we record whether a tracked brand is named. Share of answer = the share of panel prompts naming the brand, averaged across engines with equal weights. Re-measured weekly. Paid placement never affects a score.

Current release

Rules

Prompt panel

Panel version 3. A share of answer is a share of this list, so the list is frozen between releases and every measurement records the version that produced it. We do not compute a change across two different panel versions — see the archive, where such rows read «new panel» instead of a movement that did not happen.

  1. best AI writing assistant for teams
  2. top project management software for startups
  3. best CRM for small business
  4. best AI coding assistant
  5. top no-code app builder
  6. best collaborative whiteboard tool
  7. best customer support helpdesk software
  8. top data analytics platform for product teams
  9. best workflow automation tool
  10. best AI note-taking app
  11. top design tool for product teams
  12. best knowledge base software
  13. best team communication and chat app for work
  14. best marketing automation platform for b2b
  15. best applicant tracking system for hiring
  16. best tool to build internal apps and admin panels

Engines we run, and the one we dropped

This release runs 3 engines: ChatGPT, Perplexity, Gemini. A third, Google Gemini, was measured and then excluded, and saying so here matters more than the tidiness of a two-engine table. Run through our relay it returned zero mentions for every brand across the whole panel — a result that is not a finding about the market but a failure of the measurement. Publishing it would have put a column of zeros next to real numbers; dropping it silently would have been worse, because the engine count is part of what the score means.

So the rule is written into the code rather than left to judgement: a build in which any engine returns zero across the entire panel refuses to publish at all. Weights are re-normalised across the engines that did answer, never filled with an assumed value. Gemini returns to the index when the relay is fixed and it produces a measurement we can defend — and its return will be a new panel version, not a quiet edit.

Ownership disclosure

DABYTE is published by VECTORY, an AI-visibility company. This is disclosed here, in the footer of every page and in /humans.txt, because a measurement is only useful if you know who ran it. VECTORY sells measurement and advisory services; it does not sell positions in this index.

Reuse

Data is published under CC BY 4.0. Cite as: DABYTE AI Visibility Index — SaaS & AI Tools, 2026-09-07. dabyte.ai

Machine copies: /api/aiv.json, /aiv.csv, /index.md, /llms.txt.