SaaS Evaluated Find a tool

AI search citation baseline

Observed citations from sample buyer questions, with a protocol for testing visibility reports.

Prelaunch benchmark

50 complete US ChatGPT web-search answers from 50 fixed buyer questions. The answers cited 122 distinct hostnames.

Most frequently cited domains

Counts show the number of answers that cited each hostname. Each question is a separate observation.

  • peec.ai 25
  • otterly.ai 16
  • tryprofound.com 11
  • scrunch.com 10
  • help.otterly.ai 8
  • promptwatch.com 8
  • searchable.com 8
  • semrush.com 6
Download the questions and cited domains

Cross-engine sample from November 19, 2026

We sampled two buyer questions through ChatGPT web search and repeated one through the Perplexity Sonar API with web search. Each row is one answer. Counts collapse repeated source links and www variants into distinct hostnames.

EngineExact buyer questionCited domains
ChatGPT web searchWhich AI search visibility tools should a US agency compare for 10 clients, with separate workspaces and citation exports?6cuescout.com, otterly.ai, peec.ai, scrunch.com
Perplexity Sonar APIWhich AI search visibility tools should a US agency compare for 10 clients, with separate workspaces and citation exports?10aboutmartech.com, deepsmith.ai, eclincher.com, get3rd.com
ChatGPT web searchHow do Peec AI, OtterlyAI and Promptwatch compare for monitoring 50 prompts across three answer engines?3otterly.ai, peec.ai, promptwatch.com

The repeated question shared 1 cited hostnames across the two answers.

Download the prompts and source hostnames

How to test a monitoring tool

  1. Use the same wording, market, language, engine surface and time window for the observed answers and monitoring export.
  2. Record the answer, named brands and complete source URLs. Count exact page citations and brand mentions separately.
  3. Repeat over multiple runs. Compare the matching URLs first, then domain overlap and how any missing answers were represented.

This edition contains the engine observations. Vendor monitoring outputs are a separate measurement, and answers can vary across runs.