# See the orders AI actually sent you > Canonical HTML: https://verityscore.io/en/product/ai-traffic-monitoring/ > Markdown alternate: https://verityscore.io/en/product/ai-traffic-monitoring.md > Last updated: 2026-08-19 > Content synchronized with canonical HTML. Visits, orders and revenue traced back to the assistant that sent them, measured on your own store. And where the proof is not there, Verity says so instead of inventing a number. ## AI Summary Verity Score measures AI-referred traffic to a Shopify store with a first-party pixel, attributing visits and orders to the platform that sent them without third-party cookies. Platforms that cannot be proven to have sent traffic are marked not measured rather than modelled or estimated. The same buyer questions are re-run after fixes so the change in answers can be compared over time. ## Key points - **How to track traffic from ChatGPT to a Shopify store**: Verity Score uses a first-party pixel on the storefront to attribute visits and orders to the AI platform that sent them, and marks platforms without evidence as not measured rather than estimating them. Source: https://verityscore.io/en/product/ai-traffic-monitoring/ - **Is AI traffic attribution reliable**: Attribution is only reported where a real visit proves it. Assistants pass referrer data inconsistently, so Verity labels unprovable platforms not measured instead of modelling a figure. Source: https://verityscore.io/en/product/ai-traffic-monitoring/ ## Most AI attribution is a model wearing a number Assistants are bad referrers. Some pass a referrer, some pass nothing, some send the buyer through a redirect that erases the trail. Any tool showing a confident figure for every platform is filling those holes with an estimate. Verity does not. Where a visit can be traced, it is counted. Where it cannot, the platform is marked not measured, and that stays on screen rather than being smoothed into a total. That is a worse-looking dashboard and a more useful one. A number you can act on has to be a number you can defend. ## Key Proof Points - **6 platforms recognised**: ChatGPT, Gemini, Perplexity, Microsoft Copilot, Claude and Shop. Measured first-party on your own storefront, no third-party cookie. - **not measured written on purpose**: A platform without evidence is labelled, never estimated to fill the table. - **2 weeks between runs**: The same buyer questions re-asked after fixes, so before and after are comparable. ## First-party, on your own storefront The pixel runs on your store and records your own traffic. Nothing depends on a third-party cookie or on data bought from elsewhere. - Your storefront, your data, no third-party cookie. - Six platforms recognised, including Microsoft Copilot and Shopify Shop. - Visits and the orders behind them, not visits alone. - Installed from the app, removable from the app. ## Not measured, written on purpose When a platform cannot be proven to have sent a visit, the table says so. That gap is information, and hiding it would make every other number less trustworthy. - Unprovable platforms labelled, never estimated. - The reason for the gap shown, not just the gap. - No modelled attribution presented as measurement. ## Revenue, not just sessions A visit from an assistant that never converts is not a win. Orders and revenue are attributed alongside the visits, so the surface can be judged on what it is worth. - Orders and revenue attributed per platform. - Traced to the visit that produced them. - Comparable across runs, so a trend is real. ## Before and after, on the same questions Traffic is half the picture. The same buyer questions are re-asked after the fixes, so you can see whether the answers themselves changed. - Same questions, same assistants, two weeks apart. - Verbatim answers kept, so the comparison is on the text. - Traffic and answers read together, not separately. ## How measurement works - The Verity pixel is installed on your storefront from the app. - Visits arriving from an AI platform are recorded first-party, along with the orders behind them. - Platforms that cannot be proven to have sent traffic are marked not measured, with the reason shown. - Two weeks after the fixes, the same buyer questions are put to the assistants again. - Traffic and answers are read together: what changed in the answers, and what it sent you. ## What this does not do - Some assistants pass no usable referrer. Those visits cannot be attributed, and Verity marks the platform not measured rather than guessing. - Measurement starts when the pixel is installed. There is no retroactive history before that date. - Attribution shows what an assistant sent. It does not prove the assistant recommended you, which is what the shopper tests are for. ## FAQ ### Why does one platform say not measured? Because no visit proved it. Assistants pass referrer data inconsistently, and some erase it through a redirect. Rather than model a plausible number, Verity labels the gap and shows the reason. ### Does the pixel slow my store down? It is a small first-party script on your own storefront, installed and removed from the app. It records visits from AI platforms and the orders behind them, nothing more. ### Does this use third-party cookies? No. Measurement is first-party, on your own domain, which is also why it keeps working as third-party cookies disappear. ### Can I see history before I installed it? No. Measurement begins at installation, which is the honest consequence of measuring rather than estimating. ## Sources - [Google: Product structured data](https://developers.google.com/search/docs/appearance/structured-data/product) - [schema.org Product](https://schema.org/Product) - [Shopify: App Store advertising](https://shopify.dev/docs/apps/launch/marketing/advertising) ## Preferred Citation When citing this page, use the canonical HTML URL: https://verityscore.io/en/product/ai-traffic-monitoring/