#004Week 4
Issue #004 · measured · waves so far: 5 · this issue’s permanent page · all issues
Matt Edward SEO’s AI visibility scoreboard
On , ChatGPT, Claude, Perplexity and Google’s AI Overviews named Matt Edward SEO in 37 of 384 answers to 32 pre-registered questions, and cited mattedwardseo.com in 15. The questions were fixed and fingerprinted before the first measurement on . Results update weekly, flat weeks included.
Named · Week 4 · Nov 18
37 of 384
answers to the 32 target questions named Matt Edward SEO: 10%, range 7–13%.
The week before (Nov 11): 19 of 288.
Cited · Week 4
15 of 526
answers listed mattedwardseo.com among their sources: 3%, range 2–5%. Controls included.
The week before: 19 of 396.
Controls · Week 4
0 of 142
answers to the 12 control questions named him: at most 3%. No work targets these, so they show what the engines do on their own.
| Engine | Calgary: who to hire | Beyond Calgary: who to hire | Trades: how to get named | How AI picks businesses | All 4 groups |
|---|---|---|---|---|---|
| ChatGPTgpt-5.4-mini | 1 of 24 4%, range 0.7–20% | 0 of 24 at most 14% | 0 of 24 at most 14% | 0 of 24 at most 14% | 1 of 96 1%, range 0.2–6% |
| Claudeclaude-haiku-5-5 | 12 of 24 50%, range 31–69% | 3 of 24 13%, range 4–31% | 0 of 24 at most 14% | 0 of 24 at most 14% | 15 of 96 16%, range 10–24% |
| Perplexitysonar | 18 of 24 75%, range 55–88% | 3 of 24 13%, range 4–31% | 0 of 24 at most 14% | 0 of 24 at most 14% | 21 of 96 22%, range 15–31% |
| AI OverviewsGoogle results page | 0 of 24 at most 14% | 0 of 24 at most 14% | 0 of 24 at most 14% | 0 of 24 at most 14% | 0 of 96 at most 4% |
| All engines | 31 of 96 32%, range 24–42% | 6 of 96 6%, range 3–13% | 0 of 96 at most 4% | 0 of 96 at most 4% | 37 of 384 10%, range 7–13% |
01By engine
Does ChatGPT recommend Matt Edward SEO?
Rarely. In Week 4, ChatGPT (gpt-5.4-mini) named Matt Edward SEO in 1 of 96 answers, so its true share is 1%, range 0.2–6%.
Claude named him in 15 of 96 (16%, range 10–24%). Perplexity named him in 21 of 96 (22%, range 15–31%). AI Overviews named him in 0 of 96 (at most 4%). Cited share is 15 of 526 answers (3%, range 2–5%), all from Claude.
Share of answers that named Matt Edward SEO, week by week. All 32 target questions pooled; every panel runs from 0 to 50%.
ChatGPT
- at most 4% Week 0Oct 230 of 96
- at most 4% Week 1Oct 280 of 96
- at most 4% Week 2Nov 70 of 96
- at most 4% Week 3Nov 110 of 96
- 1% Week 4Nov 181 of 96
Week 4: 1 of 96
Claude
- 1% Week 0Oct 231 of 96
- at most 4% Week 1Oct 280 of 96
- 27% Week 2Nov 726 of 96
- 20% Week 3Nov 1119 of 96
- 16% Week 4Nov 1815 of 96
Week 4: 15 of 96
Perplexity
- at most 4% Week 0Oct 230 of 96
- at most 4% Week 1Oct 280 of 96
- 6% Week 2Nov 76 of 96
- at most 4% Week 3Nov 110 of 96
- 22% Week 4Nov 1821 of 96
Week 4: 21 of 96
AI Overviews
- – Week 0Oct 23not measured
- – Week 1Oct 28not measured
- – Week 2Nov 7not measured
- – Week 3Nov 11not measured
- at most 4% Week 4Nov 180 of 96
Week 4: 0 of 96
All engines together
- 0.3% Week 0Oct 231 of 288
- at most 1% Week 1Oct 280 of 288
- 11% Week 2Nov 732 of 288
- 7% Week 3Nov 1119 of 288
- 10% Week 4Nov 1837 of 384
Week 4: 37 of 384
02The field
Who does AI name for AI SEO in Calgary?
In the 72 answers to the Calgary “who to hire” questions, ChatGPT, Claude and Perplexity named 25 businesses in 6 or more answers each. The largest share was 29 of 72 (40%, range 30–52%). Matt Edward SEO was named in 31 of 72.
The names are printed as the engines wrote them, as plain text: no links and no endorsement. Only the Calgary questions are counted here.
Google’s AI Overviews showed on 17 of the same 24 Calgary searches and named Matt Edward SEO in 0 of 24; a search with no overview counts as not naming him. The sites they cited most often: agencies.semrush.com (9), bulldogmarketingagency.com (5), google.com (5), digitalestatemedia.com (4), albertarank.ca (3), konstructdigital.com (3).
| Business, as the engines wrote it | Answers naming it | Share · 95% range | |
|---|---|---|---|
| Matt Edward SEO | 29 of 72 | 40%, range 30–52% | |
| KOBZZA Agency | 21 of 72 | 29%, range 20–41% | |
| SearchBeyond | 16 of 72 | 22%, range 14–33% | |
| Creative Elements Consulting | 14 of 72 | 19%, range 12–30% | |
| Luminary Software | 11 of 72 | 15%, range 9–25% | |
| Mulder Agency | 11 of 72 | 15%, range 9–25% | |
| Visibility Drip | 11 of 72 | 15%, range 9–25% | |
| AlbertaRank | 10 of 72 | 14%, range 8–24% | |
| Duo Creative Labs | 10 of 72 | 14%, range 8–24% | |
| Scale SEO | 10 of 72 | 14%, range 8–24% | |
| Blue Ocean Interactive Marketing | 9 of 72 | 13%, range 7–22% | |
| LTL Creative | 9 of 72 | 13%, range 7–22% | |
| Unified Platforms | 9 of 72 | 13%, range 7–22% | |
| GrowME Marketing | 8 of 72 | 11%, range 6–20% | |
| Prime Digital | 8 of 72 | 11%, range 6–20% | |
| Serpcore | 8 of 72 | 11%, range 6–20% | |
| AI Visibility YYC | 6 of 72 | 8%, range 4–17% | |
| Creative Web Works | 6 of 72 | 8%, range 4–17% | |
| GO-Globe | 6 of 72 | 8%, range 4–17% | |
| LGO | 6 of 72 | 8%, range 4–17% | |
| MRC SEO Consulting | 6 of 72 | 8%, range 4–17% | |
| StyleLabs | 6 of 72 | 8%, range 4–17% | |
| ThatWare | 6 of 72 | 8%, range 4–17% | |
| ThinkProfits | 6 of 72 | 8%, range 4–17% | |
| YYC WebWorks | 6 of 72 | 8%, range 4–17% |
03Method
How is this measured?
The same 44 questions every week, fixed in advance, asked 3 times each through ChatGPT’s, Claude’s and Perplexity’s official APIs, plus Google’s AI Overviews read from Calgary search results, and scored as a share of answers with a 95% range.
- Questions
- 32 target questions in four groups of 8, plus 12 controls. Fingerprint 796e1f32…, locked
- Engines
- ChatGPT, gpt-5.4-mini, located Calgary; Claude, claude-haiku-5-5, located Calgary; Perplexity, sonar, located by country only; its API has no city setting; Google AI Overviews, Google results page, located Calgary (a search with no overview counts as not naming him)
- Sample
- 3 answers per question per engine per wave: 24 per engine per group, 526 a wave
- Named
- The answer names Matt Edward SEO, the business or its domain
- Cited
- The answer’s sources include mattedwardseo.com
- Range
- Wilson score interval, 95%. A zero is shown with its upper bound, never as a bare 0%
- Controls
- 12 control questions: 0 of 142 · at most 3%
- Cost per wave
- USD 5.83
Every change to the site and to Matt’s public profiles is dated in the change log.
04Design
Why are some question groups not targeted yet?
So each change can be told apart from the engines simply changing. Each question group gets its own start date, and the groups not yet targeted act as a before-and-after comparison for the ones that are.
| Question group | Targeted from |
|---|---|
| Calgary: who to hire | at launch |
| How AI picks businesses | launch + 2 weeks |
| Trades: how to get named | launch + 4 weeks |
| Beyond Calgary: who to hire | launch + 6 weeks |
| Controls (12 questions) | never |
The launch date is set after the second baseline wave, so the start dates above are relative to it.
05Data
Where is the raw data?
The raw answers are kept for every wave and are available on request.
The verbatim questions and answers are kept off this page on purpose, so the site never shows up for its own test questions. The cluster-level numbers on this page are also a CSV file.