Omni Impact helps you track buyer questions and understand how AI engines answer them.
Prompts, models, runs, and answers
A prompt is a question saved in your workspace. Its model settings determine which AI engines can answer it. A run is a check of that question; the answer is the response from the AI engine.
One prompt can have many runs across models and time. Inspect an individual run before assuming that its result represents every answer that engine might give.
Mentions, citations, and retrievals
These signals answer different questions.
| Signal | What to look for |
|---|---|
| Mention | The answer names your brand. |
| Citation | A source referenced in an answer. Citation rate is the share of answers that cite your brand or domain. |
| Retrieval | A source an engine found while answering your question. Finding a source does not always mean citing it. |
Use Prompt Metrics to inspect answers and Source Influence to compare source roles. Engines report sources differently, so compare like with like.
Branded and unbranded questions
A branded question names your workspace's brand. An unbranded question does not. The app offers All, Unbranded, and Branded views in Prompt Metrics.
For fictional Alpine Bars, “Is Alpine Bars gluten-free?” is branded. “Which gluten-free snack bars are good for hiking?” is unbranded. Use the latter to study discovery without inserting your brand into the question.
Your brand identity helps determine which segment a prompt belongs to. Review Data Hub when a prompt appears in an unexpected segment.
Counts and scores
Counts describe answers and their brand or source activity. Scores summarize several signals. Read each label before comparing numbers.
Home shows Overall score, Website score, and Prompt score. Prompt Metrics also displays BPI, Engagement score, and rates such as Mention rate and Citation rate.
How we measure explains each number in one line. Share of Voice is a separate market-level comparison.
Workspaces and portfolio
An organization can contain client workspaces. Select a client to work with that client's data. Portfolio (Command Center) brings clients together in an organization view. Some setup and publishing actions require a selected client rather than Portfolio.
Glossary
| Term | Meaning |
|---|---|
| AEO | Answer engine optimization: improving how a brand is discovered and represented in AI answers. |
| Brand identity | Your brand name, canonical domain, and recognized aliases in Data Hub. |
| Grounded answer | An answer that uses search sources. |
| Time window | The period selected for a results view. |
| Sample size | The number of answers or prompts included in a measurement. |