> ## Documentation Index
> Fetch the complete documentation index at: https://user-guides.demand-genius.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Perception
> How AI assistants describe your brand and your competitors: sentiment, strengths, weaknesses and the sources shaping them. What the scores mean and how to change the scope.
Perception answers the question "how does AI talk about us?" The platform asks each AI model directly about your brand and about each named competitor, runs sentiment analysis on the answers, and shows you the strengths and weaknesses it keeps returning to and the sources it cites when it does. Where [Visibility](/ai-search-intelligence/visibility) asks *whether* you appear, Perception asks *how* you are described.
## Before you start
Perception runs off the brands in **Settings > Brands**. Your default own brand is the subject; competitor brands become the benchmark. Set those up first: [Setting up your brand](/ai-search-intelligence/brands).
## Reading the dashboard
Open **Perception** under AI Search Intelligence in the sidebar.
**Overall** is your AI Sentiment Index (ASI), a weighted average across all the models. Positive values mean AI describes you favourably, negative values unfavourably. AI models are positive by default, so the index is adjusted rather than raw. Three metrics sit alongside it:
| Metric | What it tells you |
| ------------------- | ---------------------------------------------------------------------------------- |
| **Signal Strength** | How confident the analysis is in the estimate. Higher is more reliable. |
| **Conviction** | How direct and decisive the sentiment language is. Higher means stronger opinions. |
| **Inflation** | How much the expressed sentiment may overstate actual perception. Lower is better. |
**Change over Time** charts the index per model across runs. **Details** breaks the current run down by model.
### Competitors
Benchmarks your brand's sentiment vs. competiEach competitor gets the same metrics plus **Advantage**: a comparative score relative to your brand. Positive means the model views them more favourably than you.
### Strengths and weaknesses
Select a brand to see the themed strengths and weaknesses the models return, each with a sentiment score and the number of times it was mentioned across responses. This is the list to take into a positioning discussion: a weakness mentioned often with a low score is where the market narrative is working against you.
### Sources
This shows you which domains are defining the way AI views and presents the different brands in your category.
See how influential your own site is (are you the defining authority on how AI views you?) and which third party sites (reviews, comparisons, industry publications) you need to prioritise.
## When it refreshes
Runs are scheduled per workspace, weekly by default. Any change to the scope triggers a fresh run at the next scheduled pass. Progress appears in **Settings > Jobs** under Prompt Execution and Analysis Pipeline.
## Troubleshooting
* **No perception data yet.** The first run has not completed. Check **Settings > Jobs** for a Prompt Execution or Analysis Pipeline job in progress. If nothing is listed, confirm an own brand exists in Settings > Brands.
* **Budget limit reached** on a job. The workspace's daily token budget for AI calls was hit. The run resumes on the next scheduled pass; contact Demand-Genius if it recurs.
* **A competitor is missing.** Check it exists in Settings > Brands with context Competitor, or add it directly in the scope editor.
* **A model shows no data.** Not every model is enabled for every workspace. Google AI Overviews in particular is a Visibility engine and does not produce Perception results.