Brand Radar

Definition

Brand Radar is the category-defining term for AI-search brand-mention tracking: a measurement discipline that dispatches a curated set of priority queries through every major AI search engine, captures the generated responses, and reports back where the brand appears, in what context, and how it compares to competitors. Citare's product of the same name implements this discipline through a 5-stage pipeline.

Why it matters

Traditional brand monitoring tools (Brandwatch, Mention, Talkwalker, Sprout Social) watch social and news channels — X, Reddit, Instagram, news sites — for brand-name occurrences. These tools were built for social listening, not AI-search visibility. The signal they miss is the most important one in 2026: when a customer asks an AI engine "what's the best X," does the AI mention your brand.

Brand Radar fills that gap. It treats AI search engines as a new measurement surface, with its own dispatch mechanics, its own response parsing, and its own competitive frame.

The 5-stage pipeline

Citare's Brand Radar implementation follows a fixed pipeline (locked in ADR 0005):

1. Knowledge Graph (KG) — build a structured representation of the brand: products, competitors, jobs-to-be-done, customer personas, content surface. Anchors everything downstream.
2. Query Guide — Sonnet drafts a curated list of priority queries grouped by persona × intent × stage. Founder reviews and edits. The Query Guide is the per-project artifact that controls what gets measured (ADR 0007).
3. Dispatch — for each query × persona × platform combination, dispatch through a geo-seeded headless browser (Cloudflare Browser Rendering for AIO, Playwright + persona-shaped prompts for ChatGPT/Claude/Perplexity/Gemini) and capture the rendered response.
4. Parse — extract brand mentions, citation context (recommended / compared favorably / cited-as-authority / passing / unfavorably-compared), and competitor mentions from each response. Vision model (Haiku) for screenshot-based parsing of AIO; text-based parsing for chat surfaces.
5. Aggregate — roll up to brand-level metrics (citation rate, share of voice, citation position) per platform × persona × query cluster, with week-over-week deltas and alerts on largest movements.

Persona anchoring

The "persona-anchored" descriptor matters because AI search engines tailor responses to inferred user context. The same query phrased by a different persona ("best CRM for a 5-person startup" vs "best CRM for a 5,000-person enterprise") returns different recommendations. Pooled measurement across personas obscures the segment-level visibility picture that drives tactics.

Brand Radar dispatches the query × persona grid rather than just the query list. ICPs are reusable across brands (ADR 0006) with brand-variable count per project.

What Brand Radar is not

• Not social monitoring. Brandwatch and Mention watch X/Reddit/news for brand mentions. Brand Radar watches AI search responses for brand mentions. Different surfaces, different mechanics, different decisions.
• Not single-platform tracking. Tools that only measure ChatGPT (or only AIO) cover one of four indexes. The four-index reality requires platform-by-platform measurement.
• Not keyword rank tracking. Brand Radar measures whether a brand is mentioned at all in an AI response, not where the brand's pages rank. The two are related but distinct — high blue-link rank does not guarantee AI citation, and vice versa.

Operating model

For pre-customer brands, Brand Radar runs as a 5-command founder loop (CLI scripts: onboard / draft / dispatch / parse / aggregate) at ~$2/week per project. For paying customers, the same pipeline runs in production on a per-tier cadence with alerts, weekly reports, and historical comparison.

See /brand-radar for the product page and /aeo-vs-seo for how Brand Radar fits into the broader AEO discipline.