Skip to main content

​
Your main analytics hub for tracking AI search performance across prompts and competitors

​
Overview dashboard (the big picture)

Your Overview dashboard is the fastest way to understand how AI models talk about your category and where your brand shows up. It’s built from daily prompt executions (“chats”) and the entities Orvi AI extracts from them:
  • Brands mentioned (from brand_mentions)
  • Sources and citations (from source_events)
  • Position and sentiment (from per-mention mention_order and sentiment_score)
On the Overview dashboard, you’ll see four core areas:
  • Visibility graph: daily “share of voice” across brands.
  • Brands: a ranking table with Visibility, Sentiment, and Position for each brand.
  • Top Sources: which domains and URLs AI models rely on most (and how often they cite them).
  • Recent Chats: the latest AI responses created by running your prompts.

​
Dashboard filters (focus your analysis)

At the top of the Overview dashboard you can filter analytics by:

​
Brand filter

Choose your brand or a competitor to focus source and chat exploration around a specific brand context.

​
Date range (preset or custom)

By default Orvi AI shows the last 7 days, but you can:
  • pick a preset window (last 7/14/30/90 days), or
  • select a custom start/end date using the calendar.

​
Models

Filter performance by AI model (e.g. ChatGPT vs Perplexity vs Gemini) to see where you’re strong or weak.

​
Topics (Brand Topics)

Filter analytics to only include prompts belonging to selected Brand Topics. This lets you focus your analysis on a specific theme (e.g. “Email marketing” vs “Attribution”).

​
Important note about Recent Chats

The Recent Chats section is designed as a “live pulse” of your latest runs. It does not follow topic/date/model filters. The main control is a toggle to show only chats where the selected brand was mentioned.

​
Main Overview sections

​
Visibility graph

The visibility graph shows daily visibility trends for your brand and other brands that appear in AI responses for your selected time period. What to look for:
  • Trends over time (don’t overreact to single-day jumps; AI outputs vary)
  • Competitor moves (who is gaining mentions for the topics you care about)
  • Model differences (some models cite different sources and recommend different brands)
To learn the exact calculation, see:

​
Brands (ranking table)

The Brands section summarizes performance for each brand over your selected time period:
  • Visibility: % of chats where the brand is mentioned
  • Sentiment: average mention sentiment (0–100) across all brand mentions
  • Position: average mention order (lower is better)
You can click Show All to expand beyond the short “top list” and see a full ranking. To learn the exact calculations:

​
Top Sources

Top Sources shows which websites AI models rely on when answering your prompts. You’ll typically see:
  • A source type chart (how citations distribute across source categories like Editorial, Corporate, UGC, etc.)
  • A Top Domains table, including:
  • Used %: how often the domain appeared as a source across chats
  • Avg Citations: how many explicit citations the domain gets (on average) when it’s used
  • Type: the domain category/classification
Click Show All to dive into the full sources view.

​
Recent Chats

Recent Chats are the actual AI responses created when Orvi AI runs your prompts. Why they matter:
  • They explain why a metric changed (new competitor names, different ordering, new sources, tone shifts).
  • They help you identify new brands you may want to track.
  • They reveal source opportunities (domains worth targeting for PR, partnerships, or content).
By default you’ll see all recent responses, but you can toggle to show only chats that mention the selected brand. For a deeper guide, see:
  • client_facing_documentation/08-understanding-chats.md

​
Brand visibility vs source visibility (why both matter)

Orvi AI distinguishes between two different “ways you can show up”:
  • Brand visibility: your brand name is explicitly mentioned in the response (from brand_mentions).
  • Source visibility: your domain/content was used or cited, even if your brand isn’t named (from source_events).
This is why you can see patterns like:
  • High Used %, low Visibility: your content influences answers, but the model doesn’t name your brand.
  • High Visibility, low citations: the model mentions your brand, but doesn’t treat your site as a reference.
Use both to decide whether your next GEO action should be brand authority work, source/citation work, or both.
Last modified on January 25, 2026