Tracker

The cycle

1. Measure (September 2026): We asked ChatGPT, Claude, Gemini and Perplexity 2,290 questions about Atlanta and traced all 63,532 sources they cited.
2. Set the map (September, now): We publish the stories AI is missing, where it gets its information, and who can change each one.
3. Act (Through November): Programs, investors, universities, the city and the community each take their part. Everything is optional and most of it takes an hour.
4. Measure again (week of December 7, 2026): The same 340 questions, asked the same way, published whatever they show.

The numbers

• Ranked city lists that put Atlanta in the top five: 57% in September; December not yet measured
• Answers about which US hubs are rising recommend Atlanta: 0 of 40 in September; December not yet measured
• Answers about support for a fintech startup here recommend an Atlanta organization: 8 of 32 in September; December not yet measured
• Answers about raising a Series A recommend an Atlanta investor we track: 3 of 24 in September; December not yet measured
• Answers about hiring senior engineers here name no Atlanta organization at all: 6 of 24 in September; December not yet measured
• Universities AI names are never recommended by it: 3 of 8 in September; December not yet measured
• Answers to a company moving to Georgia name none of the organizations we track, the state's own agencies included: 9 of 28 in September; December not yet measured
• Mentions of an Atlanta organization becomes a recommendation: 1 in 7 in September; December not yet measured
• Places Atlanta organizations are missing from the eight lists AI cites most: 45 in September; December not yet measured

What we expect

• The lists move first. Rows added to pages the engines already cite are the only change with time to be re-read. If a third of the 45 missing places get filled, we expect those organizations to be named more, and some recommended for the first time.
• Plain beats more. Organizations that publish who they are for and point founders onward should turn mentions into recommendations faster than organizations that publish more pages.
• The city number barely moves. Atlanta is in the top five of 57% of ranked lists today. Ninety days of page edits is not enough to change what the national rankings say. We expect a small move, and we will say so plainly if there is none.
• New pages are a bet on the next cycle. A page published in November has almost no chance of being read and cited by December.

What could make it meaningless

• The engines change under us. A model update can move every number without anyone here doing anything. We will say when it happens.
• One new ranking. The engines lean on a handful of national lists, and one fresh report with good Atlanta figures could do more than everything here.
• Not knowing who acted. That is what the tracker is for: without it, a flat result cannot tell us whether the idea is wrong or whether nobody moved.
• Our own measure. "Top five" here means the share of ranked lists that put a city in the first five. Counted another way, by how many lists it appears in at all, Atlanta is second.
• Small numbers. Some of these rest on a few dozen answers, which is why the count is printed next to every figure.