Reference
Definitions, and where the numbers come from. In this chapter: Glossary, Methodology, Support.
In this chapter
Definitions, and where the numbers come from.
Glossary
| Term | Meaning |
|---|---|
| AVI | AI Visibility Index: the share of AI answers that mention your brand. |
| SOV | Share of Voice: your mentions over yours plus every tracked competitor's. |
| Citation Coverage | The share of AI answers that cite your site as a source. |
| Citation Share | Your site's share of all the sources the answers cite, averaged across models. |
| GEO | Generative Engine Optimization: optimizing content for AI-powered search. |
| AEO | Answer Engine Optimization: tuning content to land inside AI-generated answers. |
| BLUF | Bottom Line Up Front: the direct answer in the first 40 to 60 words. |
| Prompt | A question Refinea asks the AI. Generic, or tied to a persona. |
| Persona | A profile Refinea uses when asking the AI questions on your behalf. |
| Topic | A theme of your market. Prompts, metrics and Actions are grouped by topic. |
| Run | One AI answer: one prompt asked to one model in one analysis. |
| Analysis | One cycle that asks every active prompt to every model of your plan. |
| Action | A visibility problem on one topic, found by the weekly analysis, with a plan to fix it and a measured result. |
| Baseline | The metric on an Action's tracked prompts when you accepted it. Results are measured against it. |
| Workflow | An AI agent made of steps that does one piece of work, such as writing an article or auditing a page. |
| Skill | An instruction pack that shapes how an AI step of a workflow writes or judges. |
| Credit | The unit workflows are paid in. One credit is one euro cent; your plan grants a monthly balance. |
| Asset | A verified fact in Brand Memory: a proof point, an expert voice, or a fact or statistic. |
| Brand Kit | Who you are, what you offer and how you sound. Read by every workflow that writes. |
| Owner domain | A domain you control. Refinea tags citations from these as your own. |
| Ref | Refinea's assistant: it reads your data and these docs, and proposes changes you approve. |
Methodology
The metrics in Refinea come from real answers to real questions, asked to the live versions of the AI assistants. We ask the same prompts every day, so trends are comparable over time.
AI search is a young field. Some of what we measure (entity density, ideal block size, the impact of structured data) comes from peer-reviewed research; some comes from industry studies; some from practitioner heuristics. We try to be honest about which is which. When we present a number, it's a measurement. When we present a threshold, it's a reference, not a rule.
LLMs change. What works today may not work in six months. Refinea's workflows and metrics are modular for that reason: we recalibrate as the evidence shifts.
Support
Write to [email protected], or use Get help in the Help menu at the bottom of the sidebar. For a question about your numbers, Ref usually answers faster: it reads your data and these docs.