- Horizontal: reliability. The brand’s Appearance Rate, how often it appears across repeated runs of your prompts.
- Vertical: dominance. The brand’s Share when present, how much of the brand conversation it owns in the answers it’s part of.
Reading the quadrants
Inspecting a brand
Click any dot to open the brand’s detail panel:- Share of Voice, the brand’s share of every answer, absent answers as zero
- Appearance with its confidence range, so you can tell a stable position from a volatile one
- Share when present, the brand’s dominance in the answers it appears in
- Citation Rate, how often answers use the brand’s website as a source, read beside Appearance and never added to it
- Change in appearance versus the previous audit, so you can see who’s gaining or losing presence
- Share of Voice per engine, on the same basis as the brand’s headline figure, with the brands that lead Share of Voice marked on each engine’s bar. It often reveals that a competitor’s lead comes from a single engine. Share when present for each engine is shown on hover.
Filtering by page, prompt, or topic
Above the landscape you can narrow the whole view to a single page, prompt, or topic, or to prompts that do or do not name your brand. The filter applies to the landscape, the appearance distribution and the selection winners at once, so you can move from “where do we stand overall” to “where do we stand on this exact question or page” without leaving the competitive framing. The headline numbers recompute for whatever you filter to.Appearance distribution
A project’s Appearance Rate is an average across many prompts, and an average hides its own shape. The appearance distribution shows that shape: how your per-prompt appearance is spread out, so a headline like “appears in 91% of runs” reads as “we appear in nearly every run on most prompts, with a weak tail” rather than an opaque number. Group the spread by prompt, page, or topic, and a ranked view surfaces exactly where you are invisible, so you know what to fix first.Who the engines recommend
Competition prompts are the market questions with no page behind them: which providers come into question for a need. Their answers say more than who gets mentioned. They show who the engines send a buyer to. The Who the engines recommend card lists every brand the engines recommended on those questions, one row per brand and one column per engine. A cell counts the questions where that engine recommended the brand, either as the one best option or for a specific need. The last two columns split each brand’s picks into those two kinds. Both kinds count alike, because a brand recommended for a specific need is the first choice for every buyer with that need. Each engine answers a prompt several times, and one of those answers is judged, the same one the prompt’s page shows. The Prompts answered row gives each engine’s base, and a cell’s shade shows its count as a share of that base, so a shade means the same on a project with 2 competition prompts and one with 30. Engines often answer a market question without naming one best option, so the card also says how many of your competition prompts got a best pick at all. The card follows the filters above it. A page filter leaves no competition prompt, because competition prompts have no page.Movement
Once two audits (snapshots) exist, a Movement tab appears. It shows how brands shifted between audits: who gained presence, who gained dominance, and who faded. It follows your brand and the competitors you track on the Brands screen, or the three leading competitors when you track none. Movement is where you validate whether your content changes displaced a competitor, and where you catch a competitor climbing before they reach the top right.The landscape includes every brand the engines mention, not a predefined competitor list. Brands you didn’t consider competitors showing up reliably is itself a finding.