Citation share of voice across the major AI answer engines, by beat | Current posture →
Brand
URL
Queries
Competitor set
Loading data…
Share of Voice
Share of voice = C&EN citations divided by all tracked citations (C&EN plus Direct plus Adjacent competitors). Presence = share of queries where C&EN is cited or named.
Coverage Map
Most-cited sources
Across all engines and queries run so far.
Share of Voice by beat
Priority map — impact by winnability
Each cell counts queries at that impact (rows, top = highest) and winnability (columns, right = highest). Top-right is where C&EN both matters most and can realistically win. Click a cell to filter the matrix to those queries; click it again to clear.
Query matrix
Tier
Type
Impact
Winnability
Leaders by engine
Cited leaders — publishers ranked by citations, blended across the five engines by default. Use “Leaders by engine” above to re-rank for a single engine. The winner is emphasized; C&EN is maroon when it wins and outlined when it is in the field but not winning.
Engine cells — how C&EN appears in that engine’s answer:
MC mentioned & citedC cited only — linked as a source, name not in the textM mentioned only — named in the text, not linked– absent● met the win condition● partially met (Core queries only)
Winner is the most-cited publisher across engines for that query. Engine cells show C&EN's own status; a dot marks a passed win-condition judgment. Hover a query for its win condition.
Open-weight panel
Beyond the five main engines, this panel watches the open-weight models (DeepSeek, Qwen, Mistral) that power many third-party AI tools, notably in China and Europe. The two tracks ask two different questions.
Track 1 – App presence in China
Does C&EN surface inside the consumer DeepSeek and Qwen apps when a Chinese user asks about chemistry? We run 15 Core questions translated into Chinese and reviewed by a native speaker (Tiffany Chen). These apps have no API, so this is a manual check run in the apps directly, once a quarter. Hover a row for the Chinese wording.
Track 2 – Base-model authority
With no web search at all, does the raw model already treat C&EN as an authority on chemistry? We ask three “who is authoritative” questions directly to DeepSeek, Qwen, and Mistral via OpenRouter. A named result means the model volunteers C&EN from its training alone, the deepest kind of brand recognition. Runs automatically with every sweep. Hover a chip for the model’s exact wording.
Impact and winnability are starting estimates from the query set and are replaced by observed data as runs accumulate.