MODERACT · CHICAGO
AI visibility agency, Chicago.
Ask Gemini, Google AI Mode or Claude for an AI visibility agency in Chicago: across 450 answers from 2026-09-09 to 2026-09-13, 248 different businesses were named, 4 to 5 per answer.
Whether your name is one of them is what I measure for you, then fix, then measure again.
What I measure
Counts, not screenshots.
Visibility: the share of answers that name you, per engine, across many runs on several days.
Who else: the businesses that get named for the same questions.
Sources: the pages the engines cite while answering, so we know where a mention has to appear.
A screenshot is not a measurement. One answer changes from run to run and from day to day, so every number is a count across runs, with its uncertainty stated.
01 — 03
How it works
- 01You tell me the questions your buyers ask.
Five to twenty of them, with the city in the question, agreed in writing. - 02I run them, on repeat.
Each question runs 6 times a day for five days on Gemini, Google AI Mode and Claude. ChatGPT I run by hand in the app, the way a buyer would, so that column is what a buyer sees, not what an API returns. - 03You get the receipt, then the fix list.
Per-engine tables, the names that came up, the sources cited, and what I would change on and off your site, in the order I would do it.
A sample receipt
What the count looks like.
| Engine | Answers | Different businesses named | Names per answer, median |
|---|---|---|---|
| Gemini, grounded | 150 | 109 | 5 |
| Google AI Mode | 150 | 106 | 4 |
| Claude, web search | 150 | 120 | 5 |
| ChatGPT app, run by hand | 10 | 21 | 4.5 |
5 city-named questions about AI visibility agencies in Chicago, over 5 days, 2026-09-09 to 2026-09-13. Gemini, AI Mode and Claude answers are collected automatically; ChatGPT answers were run by hand in the app. Every row can be re-derived from the run log I keep.
Your receipt is the same table with your name in it: named or not, per engine, per question, with the sources each engine cited.
The next receipts show whether it worked. That is the product: the measurement, then the change, then the measurement again.
Not for sale here
What I don't sell.
Schema markup. In a matched test of 1,885 pages that added it, AI Mode and ChatGPT citations showed no significant lift, and Google AI Overviews citations dropped significantly (Ahrefs, 2026).
llms.txt files. In a log study of 137,210 domains, 97% of those files got zero requests (Ahrefs, 2026).
Rankings. I report how often you are named across many runs, never a position taken from one answer.
Promises about outcomes. I report counts; the fix list is my best judgment, and the next receipt is the check.
Pricing
$750 audit. $500 a month to keep watching.
The audit: the receipt, the fix list, and a 45-minute walkthrough. $750.
Monitoring: the same questions re-run every month, a ledger of what changed, and the next fix. $500 a month.
- 01The audit is paid when the report lands, not before.
- 02Monitoring is month to month. Stop whenever.
- 03Every number in the report can be re-derived from the run log I keep.
Who
A small Chicago team.
Led by me, Marcell Jackson, at 220 N Green St in Chicago.
Email me the questions your buyers ask and your city, and I will reply within a business day.