How We Collect Data
A2ZReach asks the AI assistants the prompts you track, saves each full answer with its citations, and reads your numbers from those saved answers.
How We Read an Answer
A2ZReach asks each engine your tracked prompts and saves the full answer with its citations and capture time. ChatGPT and Google AI Overviews are read in a browser, the way a person would; Perplexity, Gemini and Claude are read through their APIs. If a browser read fails, it falls back to the engine's API, or to a search-results provider for Google AI Overviews, and the answer records that it did. Your first report during setup reads ChatGPT and Perplexity through their APIs so it's ready in minutes. Answers vary by engine, location and time, so A2ZReach reads trends across repeated runs rather than any single answer.
The Route for Each Assistant
| Assistant | Usual route | If that fails |
|---|---|---|
| ChatGPT | A browser, the way a person would ask | The OpenAI API, with web search |
| Google AI Overviews | A browser, the way a person would search | A search-results provider that returns Google's live results |
| Perplexity | Perplexity's API | None: the API is the usual route |
| Gemini | Gemini's API, with Google Search | None: the API is the usual route |
| Claude | Claude's API, with web search | None: the API is the usual route |
When a Browser Read Falls Back
A browser read that is blocked, has no session to use, or comes back with no answer text drops to the assistant's lower route in the table, and the answer is saved with a note that it fell back. A read that is only throttled waits and tries again rather than falling back. Your first report during setup is the exception: it asks the APIs directly and never falls back, so it is ready in minutes.
Which Assistants
A2ZReach tracks ChatGPT, Gemini, Perplexity, and Google AI Overviews. The free scan reads ChatGPT; on Starter and Growth you choose 3 of the 4 tracked assistants, and Enterprise tracks 5, including Claude. Results stay separate by assistant, so you can see where you're strong and where you're missing.
How Often
It depends on your A2ZReach plan: Starter runs your prompts every 48 hours, Growth runs your prompts every day, and Enterprise runs your prompts every day. Free is a one-time scan.
Answers vary by assistant, location and time. A prompt that names your brand today might not tomorrow, so what matters is the trend across repeated runs, not any single answer.
One Prompt, One Assistant, One Day
Every prompt, assistant and day is its own captured answer. You can open any one to see what happened: was your brand named, which competitors appeared, and which sources were cited. See the AI Response Viewer for what that screen shows.
What We Read From Each Answer
- Brand mentions: whether your brand was named, and your place among the brands the answer names
- Competitors: every other brand the answer names, so you can see who is winning the prompts you care about
- Citations: the pages the answer cited as its sources, and whether a page is on your own site
- Share of Voice: your mentions compared with your competitors' across all tracked prompts
Where the Reasons Come From
The reasons and gaps in Exclusion Analysis come from a follow-up question A2ZReach asks the assistant in the same conversation, on a schedule set by your plan. Where the assistant can give one, a reason carries a quote from its own answer. Either way a reason is the assistant's account of its answer, not a measurement of it.