A

A/B testing (AI answers):

Controlled experiments that compare two versions of a passage, schema pattern, or page to see which earns more inclusions or better citations in AI answers. In Thirdeye, run the same query bank before and after a change and compare Share of Answer and SOV.

AEO: Answer Engine Optimisation:

Practice of winning inclusion inside AI answers, not only traditional SERPs. Focus areas: entity clarity, structured facts, quotable passages, trusted citations.

AI Brand Monitoring:

Tracking how your brand appears in answers and recommendations across ChatGPT, Claude, Gemini, Perplexity and similar. Thirdeye does this in real time and breaks it down by platform, sentiment and competitive context.

AI Engine Optimization (AEO):

Thirdeye’s positioning: monitor and improve how LLMs talk about your brand across platforms. Synonym of AEO: Answer Engine Optimisation.

AI Platform Breakdown:

Analysis of which platforms mention you, how often, and with what sentiment. Useful for prioritising effort.

AI Traffic:

Website visits attributed to AI answers and recommendations, including direct links from AI, post-answer searches and referral UI clicks. Track volume and conversions.

Alert fatigue:

When too many alerts cause teams to ignore them. Use frequency caps, quiet hours and escalation tiers to keep alerts actionable.

Alert rules:

The conditions that trigger notifications, such as mention spikes, sentiment drops or competitive surges. Supports multi-condition logic and dynamic or static thresholds.

Alternative prompts (custom prompts):

Queries designed to test “alternative to X” scenarios that often drive switching behaviour. Measured for mention rate and rank.

Automation & scheduling (prompts):

Run prompts daily/weekly/monthly, rotate platforms and retry on failure so results are comparable over time.

B

Brand disambiguation:

Making it explicit which entity you are. Use short descriptors, canonical URLs, schema and Wikidata so models resolve you correctly.

Brand monitoring:

Ongoing detection of direct and indirect mentions of your brand in AI answers, with context, sentiment and platform. Thirdeye supports passive and active monitoring.

Brand variations:

Alternate names, misspellings, product names and handles you include in monitoring to avoid missed mentions.

Brand visibility (AI):

How often and how well your brand appears across generative answers by model and locale.

C

Canonical page:

The definitive page for a concept or product, kept concise, structured and well-cited so models reuse it.

Citation density:

Variety and count of sources an answer uses. Higher density from reputable TLDs improves reliability and inclusion odds.

Claim verification:

Checking AI claims about your brand against your sources; useful for catching hallucinations and omissions.

Competitor alerts:

Notifications when competitor mentions surge, or when your SOV drops below a threshold.

Competitor tracking:

Monitoring rival mentions, positioning and share trends across platforms.

Confidence score (mention):

A model-side or system score for how confidently a mention was identified. Thirdeye stores confidence with context.

Content passaging:

Writing compact, self-contained paragraphs that directly answer likely questions, improving extraction.

Custom prompts:

Your own questions that simulate real buyer queries, comparisons and “alternatives”, submitted on a schedule to measure inclusion, rank and sentiment.

Custom prompt monitoring:

The workflow that runs, stores and analyses custom prompts for trends and gaps. Synonym of Custom prompts (operational view).

Custom reports:

Automated or on-demand reports with selected metrics, competitor data and commentary, delivered by email/Slack/API.

D

Dashboard (analytics):

Main workspace with executive cards, trends, platform performance, competitive intelligence and content analysis.

Disambiguation page:

A page that lists entities with similar names and clarifies which is yours to reduce name collisions.

Dynamic thresholds:

Alert thresholds that auto-adjust to historical patterns and seasonality to cut false positives.

E

E-E-A-T proxies:

Practical signals for experience, expertise, authority and trust: bios, citations, org schema, consistent naming.

Email alerts:

Notification channel for instant alerts, daily digests and weekly reports, with role-based recipients.

Entity / entity resolution:

A distinct machine-recognisable “thing” and the process of matching mentions to the right one. Strong metadata and schema improve resolution.

Escalation rules:

Step-up flows from warning to critical to emergency, with channel mix and response-time expectations.

Executive summary cards: Top metrics at a glance such as total mentions, sentiment, SOV and AI traffic.

F

FAQ blocks (AI-oriented):

Q&A sections that mirror real queries and seed answer units.

Feature attribution analysis:

Shows which features AI associates with your brand and where competitors lead.

Feedback loop (Act → Crawl):

Publish fixes, rerun the query bank, verify lift, and keep iterating.

Forecasting / predictive analytics:

Trend and sentiment forecasting, plus competitive trajectory projections.

G

GEO: Generative Engine Optimisation:

Optimising for inclusion across all generative surfaces, not only search-adjacent ones.

Geo-localisation (answers):

Inclusion patterns differ by country and language; segment tests and content accordingly.

H

Hallucination:

Confident but wrong output. Track recurring false claims and publish targeted, cited passages to correct them.

Head-to-head comparisons:

Direct comparisons of you vs key competitors across mentions, sentiment and platform performance.

High-intent query:

Queries that signal readiness to buy or switch; prioritise these in your prompt bank.

I

Inclusion rate:

Percentage of runs where your brand appears for a query set. Related to mention rate.

Integrations & webhooks:

Connect alerts and reports to Slack, email and external systems; webhooks deliver structured JSON for workflows.

Intent clustering:

Group queries by outcome: learn, compare, buy, troubleshoot.

K

KB: Knowledge base:

Public, structured facts that models can quote.

Keyword triggers:

Phrases like “best [category] for [audience]” you monitor to catch relevant AI answers.

Knowledge cutoff / knowledge graph:

Training horizon and the web of entities/relations; both affect inclusion and accuracy.

L

Linkless mention:

A reference without a hyperlink. Still moves the model’s understanding.

Locale variant:

Language or country-specific versions of queries and pages.

M

Mention analytics:

Trends in volume, distribution and correlations for brand mentions.

Mention rate:

Percentage of prompts or conversations where your brand is named. Synonym of Inclusion rate (Thirdeye uses “mention rate” for prompt performance).

Mention spike:

Short-term surge above baseline that triggers alerts; investigate cause and impact.

Mobile push:

Optional channel for instant notifications on higher plans.

Model matrix:

Grid of models and modes (browsing on/off, search vs chat, locale) for consistent testing.

N

Name collision:

Multiple entities share a name; fix with descriptors, schema and consistent metadata.

Notification channels:

Email, Slack, webhooks, dashboard, and mobile push. Configure per alert type and role.

O

Omission:

Failure to appear where you reasonably should, often due to weak entities, sparse citations or answer truncation.

Outlier claim:

A statement that deviates from known facts; triage with claim verification.

P

Passive monitoring:

Watching organic conversations and capturing unprompted mentions and recommendations. Synonym of Real-time monitoring in practice, but used for organic flow vs scheduled prompts.

Platform audience analysis:

Understanding who uses each platform and tailoring optimisation accordingly.

Platform distribution:

The mix of platforms driving your mentions. Synonym of Platform mix.

Position ranking (list rank):

Your average position when AIs list options; aim for top 3. Synonym of Average rank in list.

Prompt bank:

Canonical list of questions you test across models.

Prompt performance:

Core metrics for prompts: mention rate, rank, sentiment, SOV, response quality.

Proactive risk control:

Detect and correct inaccurate AI responses about your brand, reducing negative sentiment and misinformation.

Predictive alerts:

Alerts that anticipate trends using pattern recognition and forecasting.

Q

Query bank:

The structured set of buyer-like questions you monitor for inclusion and positioning over time.

Query intent:

The goal behind a query; match pages and passages to it.

R

RAG: Retrieval-augmented generation:

Answers assembled with live sources; your crawlability and structure affect inclusion.

Real-time monitoring:

Continuous scanning for mentions across enabled platforms with entity recognition and relevance filtering.

Recency bias index:

How often answers cite very recent material; helps plan refreshes.

Relevance score:

Score reflecting how on-topic a mention is.

Release-note monitoring:

Watching vendor updates and correlating with shifts in inclusion or stance.

Response quality:

Depth, accuracy and recommendation strength of mentions in AI answers.

S

Scheduling (alerts and prompts):

Frequencies for monitoring and notifications; start with essentials, expand as you stabilise.

Schema coverage:

Breadth and quality of schema.org on your site; improves extraction and entity resolution.

Search-grounded mode:

Assistant mode that brings in fresh web results; inclusion depends on crawlability and trust signals.

Sentiment score:

Average positivity/negativity of mentions; track trend and thresholds.

Share of Answer (SoA):

Percentage of answer surfaces that include your brand for a defined query set.

Share of Voice (SOV):

Your mentions divided by total category mentions across competitors; a core competitive KPI. Synonym: SOV.

Slack alerts:

Real-time team notifications with quick actions and deep links to mentions.

Smart alert filtering:

Context-aware filtering that predicts trend continuation and business impact, reducing noise.

Source diversity:

Spread of domains and TLDs cited in answers; stabilises narratives.

Stance analysis:

Whether an answer is favourable, neutral or critical about your brand.

Structured claim:

Machine-friendly statement with numbers, units and a citation.

System alerts:

Notices about platform status or system changes that affect monitoring.

T

Thirdeye (product):

AI-search insights platform for performance-oriented marketing teams that tracks mentions, sentiment, prompts and citations across major AI tools.

Thirdeye framework:

Crawl & Ask → Extract → Score → Compare → Alert → Act. A closed loop to measure inclusion, find gaps and verify fixes.

Topic clustering:

Grouping common themes in mentions to spot strengths, gaps and content opportunities.

Traffic monitoring (AI-referred):

Tracking visits and conversions attributed to AI answers. Synonym of AI Traffic.

Trend analysis:

Time-series view of mentions, sentiment and SOV with patterns and seasonality.

U

Unsited answer:

An AI response with no links. Publish quotable, citable passages to encourage future sited outputs.

User-perceived accuracy:

Whether the answer feels correct to readers, which affects trust and conversion.

V

Variant testing:

Trying multiple phrasings/layouts of the same claim to see which wins inclusion.

Visibility audit (AI):

A structured review of where you appear, how often and in what stance across models and locales.

W

Webhook:

A push endpoint to receive alert payloads in your systems; includes alert type, triggers and context. Synonym: Webhooks.

Widget management (dashboard):

Customise dashboard layout, show/hide metrics, resize charts and create role-specific views.

Wikidata entity:

Public unique identifier for your brand or product that helps models resolve your entity.

Win/Loss tracking (answers):

Where AI prefers you vs a competitor in side-by-side comparisons and for which features.

Z

Zero-click behaviour:

Users get the answer without clicking out, which makes inclusion inside AI answers critical.

Zero-mentions alert:

Fires when your brand isn’t mentioned for a set window; prompts checks on config, access and prompt strategy.