A new online store where ChatGPT is the largest traffic source: 80 of 163 sessions since launch
A small e-commerce built around one buyer question Brazilian search does not answer well: which Arabic perfume is the equivalent of a designer fragrance. The store is mine, launched in August 2026, and its name is withheld while it is small. It is on this page for one reason: the traffic source split is unusual and it is exported, not estimated.
- 80sessions from AI assistants, 49% of 163GA4 · 1 Jan to 21 Sep 2026
- 17countries with users in GA4, 18 key eventsGA4 · since launch
01 · ContextThe business
Arabic perfumes sold online in Brazil. Catalogue pages that state, in the first lines, what each perfume resembles, how long it lasts and what it is not.
02 · Starting pointWhere it started
Launched on 18 Aug 2026 with no audience and no ads. The GA4 property has data from launch.
03 · ChallengeThe problem to solve
Get a catalogue quoted by AI assistants that browse the web, and measure whether that produces visits and actions, on a store with no brand and no budget.
04 · StrategyWhat the plan was
A product page standard written to be quoted, structured data, a product feed and an llms.txt so ChatGPT with browsing, Google and Bing read the catalogue as a catalogue, and GA4 key events on the actions that lead to an order.
05 · StructureWhat I built
- The store: catalogue, product pages, similar-fragrance index, checkout and order tracking, on a static architecture with payment and shipping integrations.
- A product page standard written to be quoted: what the perfume resembles, how long it lasts, what it is not.
- Structured data, a product feed and an llms.txt.
- GA4 key events on the actions that lead to an order.
06 · TimelineDated milestones
- 18 Aug 2026Store launched. GA4 data starts.
- 21 Sep 2026GA4 export date (partial day).
07 · From impression to actionEach step, its own measurement
Impressions are appearances in a search or AI experience. Clicks are recorded at the source. Sessions are recorded on the site. Key events are actions the site tracks. Leads and deals live in the client's CRM. None of these is converted into another.
08 · Before and afterSame-length windows, no overlap
09 · EvolutionDrawn from the daily export, nothing smoothed
10 · GeographyWhere the demand appeared
Countries that actually visited (GA4)
17 countries with at least one active user, 1 with a key event. A GA4 user is a device or browser as the tool counts it, not a verified person.
Paraguay 2
Portugal 2
Czechia 1
France 1
Hong Kong SAR China 1
India 1
Peru 1
Puerto Rico 1
Slovenia 1
South Korea 1
Taiwan 1All countries as a table (ISO code, sorted by users)
| # | ISO | Country | Users | Key events |
|---|---|---|---|---|
| 1 | BR | Brazil | 82 | 18 |
| 2 | US | United States | 40 | 0 |
| 3 | CN | China | 5 | 0 |
| 4 | DZ | Algeria | 3 | 0 |
| 5 | DE | Germany | 3 | 0 |
| 6 | CA | Canada | 2 | 0 |
| 7 | PY | Paraguay | 2 | 0 |
| 8 | PT | Portugal | 2 | 0 |
| 9 | CZ | Czechia | 1 | 0 |
| 10 | FR | France | 1 | 0 |
| 11 | HK | Hong Kong SAR China | 1 | 0 |
| 12 | IN | India | 1 | 0 |
| 13 | PE | Peru | 1 | 0 |
| 14 | PR | Puerto Rico | 1 | 0 |
| 15 | SI | Slovenia | 1 | 0 |
| 16 | KR | South Korea | 1 | 0 |
| 17 | TW | Taiwan | 1 | 0 |
Google Analytics 4, Demographic details: Country, 1 Jan 2026 to 21 Sep 2026. 3 users had no country set. Users in data-centre cities (Ashburn, Council Bluffs, Boardman) are likely automated and are not claimed as audience.
11 · AI visibilityThree different measurements, kept apart
Traffic detected from AI platforms: 80 sessions whose recorded source is an AI assistant, chatgpt.com / ai-assistant 80. From GA4 session source/medium. This proves the origin of the visit, not the content of the conversation.
12 · TrafficWhat arrived on the site
163 sessions since launch, grouped by source family. Referral includes the previous domain where one exists. Every row is text, so the split is readable without the bars.
GA4, session source / medium, 1 Jan 2026 to 21 Sep 2026. Active users 151, new users 151, average engagement 41 s per active user.
13 · ReadingWhat changed
Of 163 sessions recorded by GA4 since launch, 80 came from chatgpt.com, 49% of the total and more than direct (71) or Google organic (11). By first user source, 75 of 151 users arrived through ChatGPT first. Users came from 17 countries, and the 18 key events recorded all came from Brazil.
14 · LimitsThe honest part
151 users is a small base and some of the direct traffic is my own testing. The Search Console export for this property was not part of the 21 Sep 2026 package, so no impression or country figure from Google is published here. Revenue and the store's name are withheld. The claim is the pattern in one exported table: an AI assistant as the first traffic source of a new store, not the size of the store.
15 · Method and sourcesHow to read these numbers
Attribution
Growth is reported as observed after the structure went live, in the period tracked. Where the property was created with the site, the baseline is the first 28 days of the property, which is stated on every comparison. No paid traffic was run on any of these sites in the window. Sales, leads and revenue belong to the client and are not published. Full method: how the numbers are measured. Machine-readable data: portfolio-data.json.
- Google Analytics 4Property guiadaperfumaria.com, 1 Jan 2026 to 21 Sep 2026 (export window; data exists from launch; last day partial). Users, sessions by source, key events, countries, cities.
- CollectionFiles downloaded on 21 Sep 2026. Search Console exports end on 19 Sep 2026 because the report lags. Processed by a script that keeps a provenance entry per figure.
Logos are the companies' own marks, shown to identify the client. Platform names identify the source of the data and imply no partnership or endorsement.
Questions people with this problem ask
Is ChatGPT traffic real e-commerce traffic?
GA4 recorded 80 sessions with chatgpt.com as the source and 18 key events on the store in the window. The sessions are measured; what the assistant said before the click is not, and is not claimed.
Why is the store's name withheld?
Because it is small and new. The data is exported and dated; the name will be published when the store is large enough for the number to matter more than the pattern.
A catalogue an AI can read is a catalogue an AI can send people to.
If your product pages cannot be quoted, they cannot be recommended. Send me one and I will show you what the assistant sees.
Run my visibility check