Terms and metrics used across Heramb.

AI Ranking Position measures where inside the AI answer your brand appears, not just whether it appears. Position matters commercially because AI-generated recommendation lists are not neutral. Users naturally pay more attention to the first few names they read. To determine rank, the system looks for explicit ordered lists or phrases like “best,” “top,” “first,” or “second.” If no clear structure exists, it uses the order in which companies are first mentioned. Being ranked first, second, or third produces meaningfully different real-world outcomes than appearing lower down, even when overall visibility numbers look similar.

Presence Rate tracks what percentage of tested prompts your brand actually appears in. Out of every relevant question asked, how many times did your company make it into the answer? Some reports call this “recommendation share,” especially when the brand appears as part of a suggested list. It’s the cleanest measure of raw inclusion. The reason it matters is straightforward: if your brand doesn’t show up, it can’t earn clicks, consideration, or trust. Many companies have solid general awareness but are completely invisible during the high-intent moments, the exact situations where people are deciding what to buy or use.

The AI Revenue Index is a directional dollar-value metric calculated as: ARI = ARS × Q × VPQ (where ARS is AI Recommendation Share, Q is Query Volume, and VPQ is Value per Query). It’s the business-friendly version of the model: how much AI-influenced demand value does your brand appear to control? It isn’t meant to be exact revenue attribution, but rather a disciplined estimate of the commercial pool tied to your AI recommendation share. It’s included so reports can speak in business language: actual estimated value, rather than stopping at abstract visibility scores.

AI Share of Voice tells you how often your brand shows up in AI-generated answers compared to your competitors. Think of it as a market-share number, but for AI responses instead of traditional search results. It answers the most basic question in AI discovery: when real users ask commercially relevant questions in tools like ChatGPT or Gemini, does your brand even appear? This metric is measured across a defined set of prompts, a specific topic cluster, a platform, or the market as a whole. It gives you a directional sense of your AI presence, but with one important catch: simply showing up is not enough. A brand can appear in answers without being recommended, trusted, or placed in a strong position. That’s why SOV should always be read alongside ranking, recommendation rate, and citation strength, never in isolation.

Attribute-level sentiment looks at how AI tools describe specific qualities of your brand: things like pricing, trustworthiness, AI features, ease of use, or customer support, rather than your brand as a whole. This matters because overall brand sentiment can look fine while specific attributes are quietly hurting you. For example, a platform might be described warmly for its user experience but negatively for pricing transparency. Strategy decisions require knowing exactly what is helping or hurting your AI recommendation quality, not just whether the brand is liked in general.

Average Rank turns an answer position into a numerical score so ranking performance can be compared across many prompts at once. The scoring works like this: first place earns 10 points, second place earns 9, and so on down to 1 point for tenth place. Any position beyond 10th scores zero. This compression is necessary because raw rank snapshots across hundreds of prompts are too fragmented to be useful. Average Rank gives a cleaner, single-number view of how strongly a brand performs across an entire cluster or platform.

Average Rank When Mentioned calculates how well a brand ranks only in the prompts where it actually appears. This isolates answer quality from answer frequency. A company might appear rarely but rank in first or second place every time it does show up. Or it might appear frequently but only in low positions. This metric separates those two scenarios and prevents analysts from mistaking an absence problem for a ranking-quality problem or vice versa.

Cluster-wide average rank measures total ranking strength across all prompts in a cluster, including the prompts where your brand doesn’t appear at all. Because it divides total ranking points by the full number of prompts in the cluster, absence genuinely hurts the score. This makes it a stronger measure of overall performance than a “when mentioned” score. It answers the harder question: not just how well you rank when you show up, but how strong you are across the complete set of opportunities available to you.

This distinction separates two types of prompts: those that explicitly name your brand (e.g., “What do users think of [Brand X]?”) versus prompts where your brand appears without being directly asked for (e.g., “What are the best tools for project management?”). The second type of organic appearance is generally more commercially valuable because it reflects an unprompted AI recommendation.

Buyer Stage identifies where a given prompt sits along the purchasing decision journey. Common stages include discovery (learning a product category exists), comparison (evaluating multiple options), and evaluation (deciding between finalists). Understanding buyer stage helps prioritize which prompt gaps matter most.

Citation architecture describes the sources and source patterns that appear to influence how AI systems talk about your brand. In other words, what websites and content types is the AI “reading” before it writes about you?

A citation moat is a durable competitive advantage that builds when your brand is consistently backed by stronger and more numerous authoritative sources than your competitors. Like a traditional moat, it’s hard for rivals to cross quickly.

Citation Source Mix shows the breakdown of source types behind AI answers: for example, what proportion comes from your own official website versus editorial articles, third-party reviews, forums, or social platforms.

Cited domains are the specific websites or root domains that show up as supporting sources inside AI responses. Monitoring which domains get cited helps you understand what external content is shaping your AI narrative.

These are the specific domains most frequently linked to your brand whenever it appears in AI answers. Mapping these domains reveals which external sources are most responsible for supporting or undermining your AI presence.

A competitive gap is the measurable difference between your brand and competitors on any key metric: visibility, ranking position, citation support, or how the brand is framed in AI answers. Gaps highlight where the most urgent optimization work needs to happen.

Competitive velocity measures how quickly competitors are gaining or losing ground in AI discovery relative to your company. A fast-moving competitor might not be ahead yet, but trajectory matters.

CPC is the estimated cost an advertiser would pay for a click in paid search on a given keyword. In this context, it serves as a proxy for commercial intent: the higher the CPC, the more businesses value that query, which signals stronger buyer interest.

Discovery Economics estimates the commercial significance of a brand’s AI visibility and recommendation performance, essentially translating AI presence into business value language.

Framing distribution classifies the role AI assigns to your brand within its answers. Common frames include leader, strong option, specialist, alternative, fallback, or cautionary option. Being framed as a “cautionary option” has very different implications than being framed as a “leader,” even if both count as appearances.

A high-intent prompt cluster is a curated group of commercially meaningful prompts that represent a specific type of buying conversation, the kinds of questions real people ask when they are actively considering a purchase or decision.

This metric shows how often a brand converts any appearance in an AI answer into a first-place recommendation. It measures conversion quality at the top of the ranking funnel.

Similar to the above, this shows how often any appearance becomes a top-three placement, a useful benchmark since users tend to seriously consider only the first few options listed.

Monthly Momentum tracks how your brand’s AI visibility metrics change from one month to the next. It’s a pulse check on whether your AI presence is growing, stable, or declining over time.

Platform Visibility compares your brand’s performance across individual AI systems such as ChatGPT, Google Gemini, Microsoft Copilot, and Google AI Overviews. Different platforms have different audiences and answer behaviors; being strong on one doesn’t guarantee strength on another.

Platform volatility measures how much performance varies either across AI platforms at a point in time or across reporting periods on the same platform. High volatility can signal unstable citation architecture or inconsistent content signals.

Prompt coverage maps which relevant user prompts a brand successfully appears in and which ones it misses entirely. Coverage gaps often represent the clearest, most actionable optimization opportunities.

Prompt subtype classification breaks a broad topic cluster into narrower prompt patterns or subtopics, allowing for more granular analysis of where a brand is strong or weak within a category.

Query intent describes what a user is actually trying to accomplish with a given prompt: learning about a topic, comparing options, checking pricing, building trust, or making a final selection. Matching your content to the right intent signals helps AI systems recommend your brand in the right moments.

Query volume measures how frequently the tracked prompts or prompt themes are being asked across AI tools. Higher volume means more commercial opportunity attached to that prompt set.

Search Volume is the estimated number of times a keyword or query set is searched within a given time period. It provides a baseline for understanding the scale of audience interest.

Sentiment measures whether the AI’s description of your brand in a given answer is positive, neutral, or negative. Sentiment can shift significantly depending on the topic, platform, or type of question being asked.

Sentiment Score and Net Sentiment summarize the overall balance of positive, neutral, and negative brand framing across all tracked prompts, giving a single number that reflects the general tone of how AI talks about your brand.

This identifies prompts where your competitors appear in AI answers but your brand does not. These are priority gaps, situations where your company should be in the conversation but isn’t.

Top-1 Rate is the percentage of all tracked prompts in which your brand is ranked first in the AI answer. It’s the clearest measure of market leadership in AI-driven discovery.

Top-10 Rate tracks how often your brand appears anywhere within the first ten ranked positions of an AI response. It acts as a broad measure of overall AI inclusion.

Top-3 Rate measures how often your brand appears among the first three recommended companies in an AI answer, a commercially important threshold, since most users give serious attention only to the top few options.

An undercontested opportunity is a prompt area or discovery segment where your company could realistically gain significant AI visibility without facing heavy competition. These represent the highest-efficiency growth opportunities in the report.

Value per Query estimates the economic value associated with a specific query or class of queries, essentially, how much revenue or pipeline a single AI recommendation in that context could be worth.

Weighted Commercial Score is a blended metric that reflects the relative business value of a cluster or prompt set by combining multiple signals, such as query volume, intent strength, and ranking position, into a single comparable score.