Weekly board
Who AI picks
How often AI answers name each brand for the same buyer questions, every week.
Top of each board
AI visibility
- ChatGPT 100 runs
- Gemini 100 runs
- Google AI Mode 92 runs
- Otterly In the mix 58% of runs 52% to 64% · 292 runs · named on 5 of 5 questions GPT 66%Gem 53%AI 55%
- Semrush In the mix 56% of runs 50% to 62% · 292 runs · named on 5 of 5 questions GPT 44%Gem 71%AI 53%
- Peec AI In the mix 41% of runs 36% to 47% · 292 runs · named on 4 of 5 questions GPT 54%Gem 36%AI 34%
- Ahrefs In the mix 38% of runs 33% to 44% · 292 runs · named on 4 of 5 questions GPT 27%Gem 42%AI 47%
- SE Ranking Long shot 19% of runs 15% to 24% · 292 runs · named on 4 of 5 questions GPT 16%Gem 18%AI 23%
- Profound Long shot 18% of runs 13% to 23% · 233 runs · named on 4 of 4 questions GPT 20%Gem 10%AI 23%
- Rankscale Rarely named 12% of runs 9% to 16% · 292 runs · named on 3 of 5 questions GPT 4%Gem 20%AI 12%
- Scrunch Rarely named 8% of runs 5% to 11% · 292 runs · named on 3 of 5 questions GPT 17%Gem 2%AI 3%
- Nightwatch Rarely named 5% of runs 3% to 8% · 292 runs · named on 3 of 5 questions GPT 1%Gem 13%AI 1%
Below the line: low confidence
Named in fewer than 10 runs, or on only one of this category's questions. Listed so you can find them, not placed.
- AthenaHQ Rarely named 10% of runs 7% to 14% · 292 runs · named on 1 of 5 questions low confidence GPT 17%Gem 10%AI 2%
- AirOps Rarely named 5% of runs 3% to 8% · 292 runs · named on 1 of 5 questions low confidence GPT 0%Gem 8%AI 7%
- BrightEdge Rarely named 2% of runs 1% to 4% · 292 runs · named on 1 of 5 questions low confidence GPT 0%Gem 0%AI 7%
- Writesonic Rarely named 2% of runs 1% to 4% · 292 runs · named on 3 of 5 questions low confidence GPT 0%Gem 1%AI 5%
Ordered by how often a brand is named across the whole category, by the low end of its 95% interval. Where a name sits in an answer is not a rank.
Movers this week
Week of October 19 so far against the week of October 12. Compared on ChatGPT, Gemini and Google AI Mode, the engines with runs in both weeks. A brand needs 20 runs in each week to be listed.
No brand cleared the mover bar: 10 points or more, on 20+ runs in each week.
Where the engines disagree
Share of each engine's runs that name the brand, last four weeks. An engine needs 10 valid runs to count; a dash means fewer.
Most-cited sources
Domains the engines cited, counted once per answer, out of 292 valid answers on this category's questions in the last four weeks.
- reddit.com Reddit · cited in 36 answers · 12% of answers
- otterly.ai Vendor site · cited in 31 answers · 11% of answers
- semrush.com Vendor site · cited in 24 answers · 8% of answers
- linkedin.com cited in 14 answers · 5% of answers
- tryprofound.com Vendor site · cited in 14 answers · 5% of answers
- peec.ai Vendor site · cited in 13 answers · 4% of answers
- nathanojaokomo.com cited in 9 answers · 3% of answers
- rankscale.ai Vendor site · cited in 9 answers · 3% of answers
- bloomiro.com cited in 8 answers · 3% of answers
- beomniscient.com cited in 6 answers · 2% of answers
Questions on the board
- What's the best tool to track whether ChatGPT recommends my clients' brands if I run a small marketing agency?
- What's the cheapest reliable tool to track my brand's mentions in ChatGPT and Perplexity?
- What are the best alternatives to Profound for a B2B SaaS company with a small marketing team?
- Which agencies are best at getting a B2B SaaS brand recommended in AI answers like ChatGPT and Google AI Mode?
- Which SEO tool is best if I also need to track Google AI Overviews and AI Mode visibility?
How we measure
Each question is asked as written, from a logged-out United States session on ChatGPT, Gemini and Google AI Mode.
How is a score calculated?
On a category board, the score is the share of every valid run across that category's questions that names the brand, over a rolling four weeks, with a 95% Wilson interval. Rows are ordered by the low end of that interval, so a brand named on one question cannot outrank one named across the category. A brand named in fewer than 10 runs, or on only one question, sits below the line as low confidence. On a question page the share is of that question's runs. Where a name sits in the answer is not a rank.
Which brands count where?
Names are matched to a reviewed brand list, and a brand counts only on questions in its own category. Names that are ordinary words, such as Streak, Loops and Instantly, count only when the answer clearly means the product. A CRM named as an integration in a meeting-notes answer is not a point on the meeting-notes board. A brand named in the question itself, as in “alternatives to Profound”, is not scored on that question, so those runs count neither for nor against it. A name that is not on the list stays off the board until a person adds it.
Will this match a signed-in chat?
A logged-out session has no memory and no personalization, so a signed-in person can see a different answer.
Cited pages are the sources that engine marked as citations, not every link in the raw response.
Run by Agnostify. Agnostify builds custom AI systems for teams, and measures how AI recommends brands. This board is a measurement, not a promise that a brand will be recommended.
Monday movers
Which names started and stopped showing up, once a week. The letter is not sending yet. This saves your place on the same list as the rest of Agnostify, tagged ai-picks. You will get a confirmation email first.