We publish our own number first.
We launched in 2026 with no case studies, and inventing some would be a strange way to sell honesty about measurement. So here is our own baseline instead — updated monthly, including the months it goes the wrong way.
Our own panel · baseline
0 / 40
Questions where an engine names Searchd today
Forty questions a brand marketer would ask before hiring an AEO agency, across five engines. Exactly what a new domain with no corpus should expect.
If it is still zero in six months, hire someone else — and this page will say so.
0 / 200observations returned an answer naming Searchd · September 2026, our first month
ChatGPT
0/40
Perplexity
0/40
Claude
0/40
Gemini
0/40
AI Overviews
0/40
- one question, asked once this month — empty means the answer did not name us
- named
5 of 5 returned an answer without SearchdSep 2026
| Buyer question | Engine | Searchd |
|---|---|---|
| AEO agency for cross-border ecommerceNamed instead: general SEO firms | ChatGPT | not named |
| how do I get my brand cited by AI | Perplexity | not named |
| how to check if ChatGPT mentions your brand | Claude | not named |
| AI visibility agency for Korean brands | Gemini | not named |
| what is answer engine optimization | AI Overviews | not named |
Five of our forty questions. Baseline recorded at launch; re-run on the first working day of each month.
The monthly report
The whole report is one page. Here is ours.
Six sections. Two are filled from our own panel with real numbers. The noise floor is a working assumption until our stability run publishes; the last three need a product and a client, so they are described — we will not invent rows to make this look better.
Searchd · monthly AI visibility
Searchd · US market · September 2026
40 questions × 5 engines
1 · Named rate
| Engine | Named | Last month | Δ |
|---|---|---|---|
| ChatGPT | 0 / 40 | — | baseline |
| Perplexity | 0 / 40 | — | baseline |
| Claude | 0 / 40 | — | baseline |
| Gemini | 0 / 40 | — | baseline |
| AI Overviews | 0 / 40 | — | baseline |
2 · Noise floor
A five-day stability run will open every engagement and set the band for that brand. Any month-over-month move inside it is reported as noise, not progress. Ours is not yet published: we are running it now, and the figure here is the working assumption of ±3 questions per engine.
3 · Named instead of you — real, from our own panel
Who came back in our place. One engine, not five: these are the 40 ChatGPT answers, categorised by what was named instead — so the rows sum to 40, unlike the per-engine named rates above, which are 40 questions each across 200 observations.
| Named instead of us | Questions (ChatGPT) | Most cited source type |
|---|---|---|
| Generalist SEO agencies | 17 / 40 | Agency directories, listicles |
| AI-visibility tracking tools | 11 / 40 | Vendor blogs, comparison posts |
| No brand named at all | 9 / 40 | Explainer articles |
| Marketing consultancies | 3 / 40 | Trade press |
Recorded at launch, September 2026, from the ChatGPT run only. Categorised rather than named because a competitor list assembled by us is not evidence — the raw transcript with the actual names is in the dataset we hand over.
4 · Factual errors stated about you — described; we have no product to get wrong
Each incorrect claim, the engine that stated it, the likely source, and whether a previously corrected error has reverted. Empty for us: nothing has been corrected yet.
5 · Assistant referral traffic — described; nothing to report at launch
Sessions and orders from chatgpt.com, perplexity.ai, claude.ai and gemini.google.com, beside your paid CPA. Understated by design — much assistant traffic arrives with no referrer — so it corroborates named rate rather than replacing it.
6 · Shipped this month — described
What we published, corrected or submitted, and what is queued for next month.
Client reporting
What a result looks like when we have one.
Case studies only with the client's named consent. No anonymous “a leading K-beauty brand” claims.
Named rate, before and after
Out of 40 questions, how many produced an answer containing the brand — by engine, at baseline and each re-run.
Screenshots with dates
The actual answer, before and after, timestamped. A number in a slide is arguable; a screenshot of ChatGPT naming the brand is not.
Corrections landed
Every factual error stated at baseline, and whether it is still being stated after remediation. In a regulated category this is the line we expect to matter most.
Judge us on your own baseline, not our case studies.
Request the free report. It measures your brand today, whether or not you ever work with us — and it is the same instrument we would be accountable to later.
- What arrives
- 20 questions, 5 engines, every answer quoted in full
- Who sends it
- Snow Lee, by hand, from Seoul
- Your email
- Used to send the report and reply about it. Privacy.
| Buyer question | Engine | Named |
|---|---|---|
| best Korean sunscreen for oily skin | ChatGPT | not named |
| K-beauty sunscreen that does not pill | Claude | named |
| Korean skincare brands sold at Ulta | AI Overviews | not named |
| …17 more questions, with the verbatim answer and every source cited | ||