What the engine labels mean
Doily is opening soon. These articles describe how the product works at launch.
Quick answer
Every AI check carries a label showing how the engine was checked: a real app, a real Google results page, a search-enabled API, or an approximation. Real ChatGPT UI and Real Gemini UI capture what the apps show. DeepSeek is labeled Approximate because it runs on open weights grounded with Google results.
Before you start
- At least one prompt added to a project
Steps
-
Find the label on each engine result
Open AI visibility and look beside each engine name. The label prints next to the numbers, together with the cadence and the location level used.
-
Read the real-interface labels
Real ChatGPT UI and Real Gemini UI mean the answer was captured from the actual app. Real Google SERP is used for Google AI Overviews, and Real Google AI Mode for Google AI Mode.
-
Read the Copilot label
Copilot Search (Bing) means the answer comes from the Copilot Search block on Bing results. It is not the Copilot chat app.
-
Read the API label
API with web search is used for Perplexity, Grok, and Claude. The answer comes from the engine's search-enabled API, so wording and sources can differ from the consumer app.
-
Read the Approximate label
Approximate (API, simulated search) is used for DeepSeek. The open-weights model runs on a U.S. cloud and is grounded with Google results, so treat its numbers as directional.
-
Check the location level
Each label also carries a location level, such as country or city, depending on the engine. A city-level check and a country-level check are not directly comparable.
Tips
- Compare an engine with itself over time. Trends on one label are more reliable than a comparison between a Real label and an Approximate one.
- Doily covers nine engines: ChatGPT, Google AI Overviews, Google AI Mode, Gemini, Perplexity, Claude, Copilot, Grok, and DeepSeek.
Troubleshooting
Why is DeepSeek labeled Approximate?
The hosted model has no web search of its own. Doily runs the open-weights model on a U.S. cloud and grounds it with Google results, which approximates the live product.
Does an API label mean the data is wrong?
No. It means the answer comes from the search-enabled API instead of the consumer app. Mentions and sources are real, but they can differ from what a person sees in the app.
Why do two engines show different location levels?
Engines accept location differently. Doily labels every check with the level it actually used instead of promising one level everywhere.
Still stuck? Email [email protected] and a person will help.
