Suadora AI visibility for law firms
Record 001 Atlanta · personal injury 24 Aug 2026 · research sample

Record 001 · research sample

Atlanta, personal injury, one morning in August.

Four questions an accident victim would type, asked of ChatGPT one after another. What came back, with its sources and its limits.

Our archive, not a client result. Firm names are on file and withheld here.


01
The record

What four runs on one morning showed.

Our archive, not a client case study. Firms are lettered here; the names are on file.

Basis of measurement
PlatformChatGPT, web interface
Date of record24 Aug 2026
Market in the questionAtlanta, Georgia
Session stateTemporary chat, memory off
Runs recorded5
Runs used4
The four questions and what came back
Question, as a person would type itSearched the webFirms named
Category. Looking for a personal injury lawyer in AtlantabYes8
Category. Best firm if not one of the two that advertise everywherebYes6
Own situation. Hit by an uninsured driver in GeorgiacYes0
Own situation. Collision with no visible damage in AtlantacYes0
Four questions, two of which returned a list4 of 414

What the answer looked likeBoth category answers came with a map and business cards, name and rating.

Question 1, cards in the answer
FirmRating
Firm A4.9
Firm B5.0
Firm C5.0
Firm D5.0
Firm E4.9
Question 2, cards in the answer
FirmRating
Firm B5.0
Firm F4.9
Firm G4.9
Firm H5.0
Firm Id4.6

a. The fifth run was discarded: with no city in the question, the assistant answered under German law from our connection. We reran it naming Atlanta.

b. Two of the six names in the second answer came from a third party ranking site, not from anything on the firm's own website.

c. Both searched the live web: the Georgia insurance commissioner and a legal directory. The claims process was explained. No firm was listed.

d. Lowest rating in either answer, on the list anyway. Picked up for the flat twenty five percent contingency fee stated on its own site.

What this record does not show

  • One engine, one metro, one practice area, one morning.
  • Each question asked once. Our threshold for stating a rate is five runs, so this is a direction, not a percentage.
  • Connection in Germany, with the city written into the question. Map cards may differ from a US address.
  • Nothing here says how many people ask these questions, and it is our archive, not a client result.

Three sources build that list: Google local signals, third party ranking pages, and the firm's own pages. Each of those is a lever.


02
The reading

Three things this record changes about the work.

The list is built from three places

Google local signals carried the map cards. A ranking site justified two names. A firm's own page put a lower rated firm on the list.

One page changed the list

Firm I had the lowest rating shown and still made it, for a flat fee stated on its own website. A page you control is an input.

The guarantee stops where the list stops

At the category question the assistant named firms. At the accident question it named none. We guarantee the first and not the second.


03
Your city

See the same four questions on your city.

We come prepared. Before the call we have already asked AI about lawyers in your city, many times over. You see who it names, who it leaves out, and why.