AI answer measurement glossary.

Position Score is a 0–100 read of how strongly AI answers place your brand in a defined market. It looks at whether models name you, recommend you, explain you clearly, trust your proof, and separate you from rivals. A Position Score is always tied to one buyer group and market: there is no single global score. The score is produced by your Association Signals read — one of the three signal categories behind every Trendscoded read, alongside Capability Signals and Narrative Signals.

For a Series B fintech selling to mid-market CFOs in North America: Position Score on ChatGPT 73, Gemini 41, Claude 58, Grok 12. The 12 on Grok flags an answer-share gap with two rivals; the 73 on ChatGPT is the position to defend.

The Strategic AEO Plan (Strategic Answer Engine Optimization Plan) is the action plan Trendscoded builds from your market read. It delivers one plan with a ranked list of 30+ next steps — close a gap, defend a strength, amplify a signal — with the proof to create. It is the marketing team's working document, replanned whenever your targets shift — not a dashboard.

Week of May 2: Gap — Grok is naming Rival X over you for 'best procurement automation for mid-market.' Defend — ChatGPT consistently names you for 'AP automation' since April. Ship — publish a third-party-cited integration case study with NetSuite to lift answer share for that comparison question.

AI Market Signal Intelligence is Trendscoded's differentiating layer above AEO visibility tracking. Visibility tools show where a brand is mentioned, cited, or ranked. AI Market Signal Intelligence interprets the directional signals behind that read — proof structures, competitor gaps, narrative shifts, third-party validation, and evidence gaps — and turns them, on demand, into a Strategic AEO Plan. It is the intelligence layer that tells a marketing team what to ship next, not just whether the brand appeared.

A visibility dashboard shows your brand dropped from third to fifth on Gemini for a buyer query. AI Market Signal Intelligence explains why: a rival gained third-party proof for a capability claim, your evidence is still first-party only, and the on-demand Strategic AEO Plan should ship an analyst-cited comparison page plus a customer proof artifact.

Trends Desk tracks the weekly trends that can move your brand's position in AI answers. It reads the proof your team ships, the market signals your brand earns, what rivals publish, and the analysts or experts shaping the category. Each item is checked for strength and fit, then tied to the trend it supports. The output is the weekly read of where the market is moving and which trends belong in the next Strategic AEO Plan.

Week of May 11: Gartner published 'federated identity adoption up 52% YoY.' Wiz published an AWS audit case study. A buyer newsletter cited your brand replacing Splunk. Your team shipped a comparison page for that buying criterion. Together, those signals support one trend — 'AI-native security stacks' is emerging as a new category — and feed the week's Strategic AEO Plan.

Mention Share is the share of AI answers — across ChatGPT, Gemini, Claude, and Grok — that name your brand for comparison questions in your defined market. It is measured over a 30-day rolling window because individual AI answers rotate. Mention Share answers the question 'do AI assistants know we exist for this buyer?' before Answer Share answers 'do they recommend us?'

Across 240 prompt-runs in the last 30 days for 'enterprise developer security tools, North America,' your brand was named in 42% of answers (101/240). Two rivals sat at 71% and 58%; the gap to close on mention is roughly 30 percentage points.

Answer Share is the conditional measure that follows Mention Share: of the AI answers that named you, how often were you placed in the top three recommendations? It captures whether AI assistants treat you as a leading option for the buyer, not a long-tail mention. Answer Share is measured over a 30-day rolling window across the four major models and reads as a percentage of mentioned answers, not all answers.

Of the 101 answers that named your brand in the last 30 days, 38 placed you in the top three (38%). Rival X's Answer Share over the same window was 64%; that 26-point gap is the lever the Strategic AEO Plan attacks first.

Capability Signals are one of the three signal categories behind every Trendscoded read, alongside Narrative Signals and Association Signals. They compare the capabilities a brand is known for in the selected market: when a buyer asks an AI model to rank vendors on the things that decide the deal, do you appear, and where? Every capability read is tied to a defined buyer frame and market — the same questions your buyers actually ask.

A Series B data-security vendor selling to enterprise IT: ChatGPT ranks it top-3 for audit-trail coverage but omits it entirely for large-fleet rollout — a capability gap named, with the proof artifact to close it.

Narrative Signals are one of the three signal categories behind every Trendscoded read, alongside Capability Signals and Association Signals. They compare whether the brand shows up in the narratives that matter — recommendation lists, comparison discussions, 'alternatives to X' shortlists — and what role it plays when it does. Where Mention Share counts how often you appear, Narrative Signals read which narratives you appear in.

For 'alternatives to the category leader' across Gemini and Grok: the brand appears in 2 of 10 shortlist narratives; two rivals appear in 7. The gap names the comparison pages and third-party citations to build next.

Association Signals are one of the three signal categories behind every Trendscoded read, alongside Capability Signals and Narrative Signals. They compare what models associate the brand with: does it come up at all for the defined market, which associations does it own, is it recommended, and how deep does the model's knowledge go? An Association Signals scan is the deep read that produces your Position Score — which is why the score is always tied to one buyer group and market.

A challenger CRM for mid-market manufacturing: Claude attaches 'field-service scheduling' to the brand unprompted — an owned association — but files it under generic CRM for everything else. That category-fit gap is what the next plan targets.

Signal Owner is the rank-1 position in Trendscoded's competitive ranking for a defined buyer frame and use case. The Signal Owner is the brand AI engines default to when constructing answers about the category — named first, cited with confidence, and used as the benchmark against which all other brands are compared. Signal Ownership is won through accumulated corroborated proof across multiple sources and sustained by a consistent weekly cadence. It is not permanent: Signal Ownership shifts as challengers build evidence and as model updates recalibrate what they know. The incumbent often holds Signal Owner by default at training time without ever having explicitly built for it.

For 'best AP automation for mid-market fintech,' ChatGPT, Gemini, and Grok all name Vendor X first, cite their case studies and G2 reviews, and use them as the comparison baseline. Vendor X is the Signal Owner for this buyer frame. A competing brand ranked 4th is in Challenger position — visible on specific prompts, absent from the category-level answer.

Leader is the rank 2–3 position in Trendscoded's competitive ranking. Leaders are named consistently in AI answers but not as the first choice — they appear after the Signal Owner and before Challengers. They have strong proof coverage on specific surfaces but haven't accumulated the breadth of corroboration required to displace the Signal Owner. The Leader position is strategically valuable: enough AI answer presence to close deals, enough gap to the Signal Owner to have a clear build target. A brand moves from Challenger to Leader by winning the specific comparison surfaces the Signal Owner doesn't dominate.

For 'enterprise procurement automation,' Rival X is the Signal Owner (rank 1). Your brand ranks 3rd — a Leader. ChatGPT names you in 7 of 10 answers but ranks you behind Rival X and Rival Y. The Strategic AEO Plan targets the two comparison surfaces where Rival Y beats you to push your brand into rank 2.

Challenger is the rank 4–5 position in Trendscoded's competitive ranking. Challengers are named in AI answers for specific queries — usually narrow use cases or direct comparison questions — but are absent from general category queries and top-3 consideration sets. The Challenger position is the most actionable: enough proof to be visible, enough gap to the Leaders to have specific surfaces to close. Trendscoded helps teams move from Challenger to Leader, then Signal Owner, by closing comparison surfaces one week at a time.

For 'best inline AI usage control,' your brand ranks 5th. You appear in 31% of answers, mostly when the prompt specifically names your use case. On the category-level prompt 'best AI security tool for enterprise,' you are absent. First Strategic AEO Plan priority: own the top two comparison surfaces where you're invisible.

Peripheral mention is the rank 6+ position in Trendscoded's competitive ranking. Peripherally mentioned brands appear in AI answers but without confidence or consistent recommendation — a passing reference in a listicle, an afterthought in a comparison. AI engines are acknowledging their existence, not recommending them. A peripheral mention is meaningfully different from Challenger or Leader because it provides no real pipeline influence: buyers running AI-assisted research don't shortlist peripherally mentioned brands. The first build priority for peripheral-mention brands is closing the gap to Challenger — winning two or three specific comparison surfaces with corroborated proof.

Your brand appears in 6% of answers for your target buyer frame — usually as the 7th or 8th item in a long list, with no cited proof. Position Score: 12. The buying committee sees you but doesn't shortlist you. First Strategic AEO Plan move: own the '[Signal Owner] vs [Your Brand]' comparison surface before building anything else.

AEO — Answer Engine Optimization — is the working name for the practice of improving how answer engines (ChatGPT, Gemini, Claude, and Grok) name, explain, cite, and recommend your brand. AEO operates inside generated answers, not blue links: the work is to make sure a model names you, understands your fit, trusts your proof, and recommends you for the right buyer.

A Strategic AEO Plan delivers a ranked list of 30+ next steps: gaps to close in answer share, strengths to defend, proof signals to publish. The plan is the weekly work list for improving how AI answers place your brand.

GEO — Generative Engine Optimization — and AEO refer to the same discipline. GEO is the term that gained traction in academic and AI-research circles; AEO is the term most marketing teams use because it parallels SEO. Both name the practice of measuring and improving how generative AI engines name, rank, and cite a brand. Trendscoded uses 'AEO' because the buyer's mental model is 'this is what comes after SEO.' The terms are interchangeable; the discipline is one.

If your team's marketing operating system says 'we run SEO, paid, and content,' add 'AEO' (or 'GEO') as the fourth engine. It owns the daily read and plan for AI-answer visibility.

Answer engines are the surface AEO works on. Where a search engine returned ten blue links and let the user pick, an answer engine returns one synthesized response that names some vendors, ranks them, cites a few sources, and skips the rest entirely. The four major answer engines Trendscoded measures across — ChatGPT, Gemini, Claude, and Grok — each behave differently: different training data, different web-grounding behavior, different citation patterns. A brand can be top-three on one engine and unmentioned on another for the same buyer query.

Asked 'best procurement automation for mid-market manufacturers,' Grok returns a four-vendor recommendation with citations to G2 and Capterra. ChatGPT returns a five-vendor narrative with no citations. Gemini returns a three-vendor list and recommends one outright. Same buyer query, three different positions for your brand.

Read the Market · Build the Proof · Strengthen your Position · Compound the Gains is the canonical Trendscoded weekly loop. READ THE MARKET is the Trends Desk reading the weekly trends moving AI-answer position across ChatGPT, Gemini, Claude, and Grok. BUILD THE PROOF is the Strategic AEO Plan: a ranked list of 30+ next steps and the proof to create. STRENGTHEN YOUR POSITION is publishing that proof and improving where you stand against rivals by buyer, use case, and model. COMPOUND THE GAINS is tracking how each plan moves Position Score, Mention Share, and Answer Share over time. The cadence is intentionally weekly because AI-answer movement is daily-noisy but weekly-meaningful.

Six weeks and six plans in: the team has shipped six Strategic AEO Plans. Position Score on Grok has moved from 23 to 47. Mention Share for a priority buyer group is up 18 percentage points. The compound effect is visible in the trend, not in any single read.

The Signal Pilot is the founder-led 24-hour introduction to Trendscoded. $500 fixed price, no subscription, no auto-renewal. The founder configures comparison questions on a 30-minute kickoff call on the model of your choice (ChatGPT, Gemini, Claude, or Grok). Within 24 hours you receive the Baseline Trends Desk read against your rivals, a Position Score read, and your first Strategic AEO Plan — a ranked list of 30+ next steps and the proof to create. A 15-minute review call at hour 24 closes the loop. The pilot is capped at the first 15 teams and is fixed price because engine costs are sunk on kickoff. After the pilot, teams either keep going plan-by-plan on the $99 Agentic plan — or take the read and walk.

Series B fintech pilot: 30-min founder kickoff configures comparison questions across two buyer groups. Within 24 hours the Baseline Trends Desk read surfaces that Rival X is gaining rank on Gemini. The first Strategic AEO Plan closes that gap by publishing a third-party-cited integration case study. The team rolls straight into its next Agentic plan.

First-party sources are the canonical attribution unit for AEO. When a claim like 'Vendor X supports SOC2 Type II' appears in an AI answer, the cleanest citation is Vendor X's own trust page — not a competitor's listicle that recaps Vendor X. AI engines cite whatever they retrieve, but first-party sources carry more E-E-A-T weight (expertise, authoritativeness, trustworthiness) than aggregator pages. The Trendscoded retrieval pipeline is tuned to prefer first-party sources for vendor-specific claims and fall back to aggregator pages only for category-level summaries.

Asked about Vendor X's enterprise readiness, an AI engine cites Vendor X's '/security' product doc as the primary source — not a competitor's '11 Best Tools for 2026' listicle that mentioned Vendor X in passing.

Aggregator pages — titles like 'Best 11 Tools for 2026' or 'Vendor X vs Vendor Y' — are content shapes AI engines lean on when summarizing categories. They're high-relevance for category-level answers and low-trust for individual vendor claims. The failure mode is when an answer engine cites an aggregator's listicle as the source for a fact about a competitor mentioned inside that listicle — a form of citation laundering. Trendscoded treats aggregator pages as weak evidence for vendor-specific facts.

An AI answer says 'Vendor X has 100+ integrations' and cites Aggregator Y's listicle. Aggregator Y has no integration data — they pulled the claim from Vendor X's own page. The cleaner citation is Vendor X's '/integrations' page directly.

Citation laundering is a structural failure mode of AI-answer citation. An aggregator publishes a listicle that recaps facts about many vendors. The AI engine retrieves the listicle, mines a fact about Vendor X from inside it, and cites the aggregator — not Vendor X's own page. The result: Vendor X's authoritative claim gets attributed to a competitor or an unrelated publisher. Citation laundering inflates the perceived authority of aggregators, dilutes first-party brand equity, and makes it harder for marketing teams to trace which content is moving AI-answer rankings.

An AI engine cites Vendor A's '11 Best AI Security Tools' listicle as the source for the claim 'Vendor B supports SOC2.' Vendor B's own trust page says exactly the same thing — the AI engine just laundered the citation through Vendor A's content.

Listicle dilution is what happens when one aggregator's '11 Best [Category] Tools' page becomes the canonical AI-answer shape for an entire market. The engine retrieves that listicle, treats its ranking as authoritative, and reproduces it across hundreds of buyer queries. Vendors not on the listicle disappear; vendors low on the listicle stay low; the listicle's editorial hand becomes the de facto market consensus. Defensive AEO move: publish your own canonical listicle in your voice, naming yourself fairly first, before competitors' versions cement.

Asked any 'best AI usage control tool' query, ChatGPT and Grok both return a five-vendor list that is structurally identical to a single competitor's '11 Best Tools for 2026' listicle. The listicle has set the rank order across the entire category until something dilutes it.

In-window evidence is content with a publication date that falls inside the recency window Trendscoded is reading. The default Trends read uses last_30_days, last_90_days, or last_180_days windows depending on category velocity. Evidence published inside the window is treated as current; evidence outside is downgraded. AI engines themselves apply similar recency weighting — a March 2024 analyst report is heavier in answer generation than a March 2022 one for fast-moving categories. In-window evidence is one way the Trends Desk tracks what changed week over week.

For a 30-day window ending May 6: an integration release announcement from April 18 is in-window. A vendor white paper from January 2024 is not. Both can be cited, but only the April release is treated as 'current evidence' in the Strategic AEO Plan.

Evergreen evidence is content that supports a trend or claim but has no clear publication date — most product pages, '/security' or '/integrations' tabs, technical documentation, API references, standards documents. AI engines treat evergreen evidence as supporting context, not as freshness signal. Trendscoded treats evergreen evidence differently from dated evidence so useful product proof is not penalized for lacking a publication date. Evergreen and in-window evidence answer different questions: evergreen establishes capability, in-window establishes momentum.

Vendor X's '/security' page has no publication date but lists current SOC2 Type II, ISO 27001, HIPAA. That is evergreen evidence of capability. A March 2026 customer case study quoting a CISO citing those same certifications is in-window evidence of momentum.

Signal Strength Score reads how well a single piece of content (a blog post, a release note, a benchmark) supports a specific trend. A strong signal directly proves the trend with named actors and concrete proof. A weak signal is only loosely related. Signal Strength is paired with Knowledge Basis: high Signal + low Knowledge means an emerging trend; high Signal + high Knowledge means a well-established trend.

A March 2026 Forrester report titled 'The Inline AI Usage Controls Wave' that names eight vendors and ranks the market scores Signal Strength 0.92. A January 2024 vendor blog post mentioning AI controls in passing scores 0.45.

Knowledge Basis Score reads how much supporting evidence exists behind a trend. It reflects how much corroborating content (analyst reports, customer case studies, vendor releases, regulatory guidance) supports the same trend. Knowledge Basis pairs with Signal Strength: a trend can be sharply signaled by one strong report but thinly known elsewhere, or weakly signaled but widely mentioned across many sources. The Signal Evidence Gap is the practical lever.

A trend like 'AI prompt firewalls embedded in execution paths' might have Signal Strength 0.85 (one Forrester report names it sharply) but Knowledge Basis 0.40 (only three vendor blog posts and one Reddit thread back it). That gap of 0.45 is the proof-signal opportunity to publish into.

Signal Evidence Gap is a derived measurement: signal_strength_score minus knowledge_basis_score. It runs −1.0 to +1.0 and points to specific marketing actions. A positive gap (signal > knowledge) means a trend is sharply signaled but under-corroborated — opportunity to publish proof and own the narrative. A negative gap (knowledge > signal) means evidence exists but no clean signal — opportunity to ship a strong analyst-style read or release announcement that crystallizes the pattern. A near-zero gap means the trend is mature and you're competing on share, not framing.

A trend with Signal 0.85 and Knowledge Basis 0.40 → gap +0.45. The Strategic AEO Plan: ship a third-party-cited customer story this week to corroborate the trend before competitors do.

Signal Stacking is the practice of building corroborated proof on multiple independent surfaces — analyst coverage, third-party reviews, customer case studies, comparison pages, practitioner threads, technical benchmarks — so that AI models encounter your brand's evidence from multiple paths when constructing an answer. A single strong case study is a signal. The same case study cited by an analyst, referenced in a G2 review, and linked from a practitioner's X thread is a stacked signal: the model encounters four independent sources pointing to the same claim. Signal Stacking is why the Strategic AEO Plan ships many next steps rather than one — position movement requires breadth of corroboration, not depth on a single surface. Stacked signals are also more resilient: a model update that de-weights one source type doesn't erase the position if five other source types carry the same claim.

Week 3 Strategic AEO Plan: publish a customer case study (signal 1), brief a practitioner to thread about it on X (signal 2), submit it for a G2 review citation (signal 3), and reference it in your comparison page (signal 4). All four sources point to the same claim from independent retrieval paths. The model encounters the claim four ways. That's signal stacking — and it's why that claim becomes harder for rivals to displace.

Tribal Context defines whose answer share is being measured. It includes the primary brand (your company), the competitor brands (named products you compete with), and the competitor vendors (the parent companies). Trendscoded uses that context to read relative position: when ChatGPT names three vendors for a buyer query, did it name your brand or someone else's? Position Score, Mention Share, and Answer Share are all scoped to this context. Without it, there is no useful 'absolute' AI-answer position.

Tribal context for a Series B procurement automation startup: primary_brand = 'Coupa challenger', competitor_brands = ['Ramp', 'Brex', 'AirBase'], competitor_vendors = ['Ramp Inc', 'Brex Inc', 'AirBase Inc']. Mention Share is now scoped to: 'in answers about mid-market procurement automation, what % named our tribe vs. theirs?'

Buyer Frame is the context for every read. AI answers vary dramatically by what kind of organization is asking and what market it operates in. A read for 'enterprise SaaS in fintech' returns a different vendor list than 'mid-market manufacturers in industrial automation.' Position Score, Mention Share, and Answer Share are always scoped to a Buyer Frame — there is no global score. Most teams run two to four buyer frames in parallel, one per buyer group they sell into.

Buyer Frame: 'Series B–D SaaS companies buying AI infrastructure.' Trendscoded runs provider comparison questions framed by this context across all four answer engines.

Market Boundary is the explicit definition of what counts as 'inside' your defined market and what counts as 'adjacent.' For a market like 'inline AI usage controls,' the boundary excludes post-use AI monitoring, network-edge SSE, browser-only tools, and standalone DLP. The boundary is critical because AI engines often blur adjacent categories — without a clear boundary, Trendscoded can't distinguish 'we lost rank in our market' from 'we got compared against an adjacent market we don't compete in.' Every comparison question is filtered through the market boundary so reads stay scoped.

Market Boundary for inline AI usage controls — what_it_is: 'execution-path control of outgoing prompts at moment of send.' what_it_is_not: 'post-use AI monitoring, SSE, browser-only controls, standalone DLP unless inline.' Adjacent markets are listed for exclusion context, not as competitors.

Adjacent Markets are categories that share buyer language with your defined market but are structurally different. For 'inline AI usage controls,' adjacent markets include SSE/SWG, browser-based AI controls, post-use AI monitoring, standalone DLP, and AI gateway routing. Trendscoded tracks adjacent markets as exclusion context: if an AI engine confuses your market with an adjacent one, the read flags it. Adjacent markets are not competitors — they are reference categories that help disambiguate the boundary.

When ChatGPT recommends a network-edge SSE tool in answer to 'best inline AI usage control,' the read flags 'boundary confusion with adjacent market = SSE.' That is the Strategic AEO Plan's lever: ship content that sharpens the inline / network-edge distinction.

Theme Ranking shows which trends are shaping the market. Where Position Score reads 'how is your brand placed?', Theme Ranking reads 'what trends are changing buyer expectations?' A Theme Ranking returns the top themes for a market, buyer, and time window: 'send-time prevention for sensitive data exposure', 'policy enforcement embedded in AI execution paths,' and similar shifts. Themes feed into the Strategic AEO Plan: which trend should this brand attach itself to this week?

Theme Ranking for inline AI usage controls, May 2026, top_n=5: (1) Send-time prevention for sensitive data exposure (rising, momentum 4.5/5); (2) Policy enforcement embedded in execution paths (rising, 4.0); (3) AI agent governance and tool-call control (rising, 3.5); (4) Browser-isolation alternatives (declining, 2.5); (5) Network-edge AI controls converging (stable, 3.0).

Question-level Visibility is the granular read beneath aggregate metrics. Mention Share at 42% averaged across 240 comparison-question reads hides which questions are at 80% and which are at 5%. Question-level visibility breaks the average open: the questions where you're consistently named, the questions where you're invisible, and the questions where you flip in and out. The Trends Desk surfaces question-level changes so the team can see exactly which comparison language is shifting.

Aggregate Mention Share: 42% on Grok. Prompt-level: 'best AP automation for mid-market': 78% mentioned. 'AP automation for retailers under 500 employees': 8% mentioned. The retailers prompt is the gap — sized accurately at the prompt level, not the category level.