Week 3 · Issue #003
On , ChatGPT, Claude and Perplexity named Matt Edward SEO in 19 of 288 answers to 32 pre-registered questions, and cited mattedwardseo.com in 19 of 396 answers.
01The numbers
How often did each engine name Matt Edward SEO?
In 19 of 288 answers to the target questions, so the true share is 7%, range 4–10%. Each cell below is 24 answers: 3 per question, asked through the engine’s official API, with a 95% range.
The 12 control questions, which the site never targets, named him in 0 of 108 answers (at most 3%).
| Question group | ChatGPT | Claude | Perplexity | All engines |
|---|---|---|---|---|
| Calgary: who to hire | 0 of 24 (at most 14%) | 19 of 24 (79%, range 60–91%) | 0 of 24 (at most 14%) | 19 of 72 (26%, range 18–38%) |
| Beyond Calgary: who to hire | 0 of 24 (at most 14%) | 0 of 24 (at most 14%) | 0 of 24 (at most 14%) | 0 of 72 (at most 5%) |
| Trades: how to get named | 0 of 24 (at most 14%) | 0 of 24 (at most 14%) | 0 of 24 (at most 14%) | 0 of 72 (at most 5%) |
| How AI picks businesses | 0 of 24 (at most 14%) | 0 of 24 (at most 14%) | 0 of 24 (at most 14%) | 0 of 72 (at most 5%) |
| All target questions | 0 of 96 (at most 4%) | 19 of 96 (20%, range 13–29%) | 0 of 96 (at most 4%) | 19 of 288 (7%, range 4–10%) |
02Changes
What changed before this issue?
Every change made before this wave is dated in the change log, with the last wave before it.
A flat week counts too: it stays on the record like any other.
03Data
How was this measured?
With the same questions every week, fixed and fingerprinted (796e1f32…) on . The full method is on the scoreboard.
These numbers are also a CSV file. The raw answers are kept for every wave and are available on request.