Pages tagged "glossary"

90 pages tagged with glossary. ← all tags

• Engaged-View Conversions (EVC): Why YouTube ROAS Reads Too High EVC credits a conversion when someone watches a YouTube ad without clicking, then buys within the window. It sits inside the main Conversions column — the reason YouTube reports can show more conversions than clicks and inflated ROAS.
• Google Performance Max (PMax): The AI Campaign That Runs Across All of Google PMax is Google's goal-based, AI-run campaign: assets + a goal in, ML-controlled bidding/placement/creative across all Google inventory out. The Google twin of Advantage+.
• Query Fan-Out — Why AI Search Doesn't Search What the User Typed AI search engines rewrite the user's question into multiple synthetic sub-queries and synthesize the answer from those. Google documents it for AI Mode; ChatGPT fans out on ~2/3 of prompts. Consequence: you can rank #1 for the typed phrase and still be invisible — the model searched a reformulated version.
• Share of Model — The AI-Search Competitive Dimension Share of Model measures how often AI systems (ChatGPT, Claude, Gemini, Perplexity) reference your brand vs. competitors when answering category questions. Parallel to Share of Voice at the traditional-media layer. The competitive dimension that didn't exist in 2024 but is now material: ChatGPT holds ~79% of generative AI web traffic; AI Overviews appear in 89% of brand searches; sector concentration is extreme (Apple = 54.38% mention share in consumer electronics). For non-dominant brands, specialization is the only path.
• FTC AI Marketing Enforcement — Deceptive AI Claims (Cox Media 'Active Listening') FTC AI-marketing enforcement under §5: the May 2026 Cox Media 'Active Listening' settlement (claiming AI you don't have is deceptive), plus the July 2026 proposed AI-accuracy policy statement — with its implied-preemption challenge to state AI laws like Colorado's.
• Agent Payment Protocols — The Infrastructure Under Agentic Commerce Four standards shipped in 2025-2026 define how AI agents transact: AP2 (Google's universal payments rail, September 2025), x402 (Coinbase/Cloudflare HTTP-native stablecoin protocol), UCP (Universal Commerce Protocol — the commerce schema), and Visa TAP (agent identity/verification). AWS Bedrock AgentCore Payments shipped May 7, 2026, making x402 hyperscaler-native. The infrastructure layer for agentic commerce is now production. The protocols are layered, not competing in most cases.
• AI Agent Behavior — What It Means The emerging field studying how AI agents make decisions, including purchasing biases. Connects to the jagged-frontier and recognition-primed-decision foundations: agent biases are predictable from the same conditions that predict human pattern-matching reliability.
• Agentic Memory — How AI Agents Remember Across Sessions Agentic memory is the architecture that lets AI agents retain useful information across sessions, tasks, and time. The category covers everything from simple conversation history to learned skills, persistent project context, and cross-session preference accumulation. Memory is the difference between an agent that re-explains itself every time and one that compounds context over weeks. The discipline of designing memory well is part of agent engineering. Anthropic's Dreaming (May 2026) is the first platform-level realization of automated semantic+procedural memory consolidation.
• AI for Competitive & Strategic Analysis — What It Can and Can't Do Can you hand competitive and strategic analysis to an LLM? Peer-reviewed evidence says yes, but only with three disciplines: aggregate many runs (a single LLM is biased), keep humans on the judgment dimensions, and let AI triage signal — not make the call.
• AI Task Restructuring — Why Idea Generation and Editing Become the New Bottleneck Noy & Zhang (2023, Science, n=444) found that ChatGPT didn't just speed writing tasks up — it changed which sub-tasks were the leverage points. Rough-drafting compressed; idea generation and editing became where humans add value. The implication: AI shifts the bottleneck, doesn't just lift it.
• AI UGC Ads: Definition, How They Work, and the Honest Case for Them AI UGC ads mimic user-generated content style using AI avatars, voice, and visuals instead of real creators — cutting cost per creative ~10x and unlocking multi-language scale, with an honest trust caveat.
• App Tracking Transparency (ATT): The iOS Privacy Change That Reshaped Paid Media ATT is Apple's iOS 14.5 privacy framework requiring apps to ask before tracking. Most users said no — collapsing ad signal and making creative the new lever.
• Content Provenance & AI Disclosure — C2PA, SynthID, and the Law Three layers that get conflated: technical provenance (C2PA metadata, SynthID watermark), platform AI labels, and disclosure law (EU AI Act, FTC, state). What each is and isn't.
• Context Engineering — Designing Information Flow for AI Agents How to structure tool responses so AI agents can reason effectively across multiple calls
• Creative Formula vs Creative Skin: The Key Distinction The creative formula is the ad's reusable structural recipe. The skin is the swappable surface — product, brand, wording. Preserve one, swap the other.
• Creative Is the New Targeting — Why Performance Marketing Leverage Moved From Audience to Creative Algorithm-driven ad platforms (Meta Advantage+, Google PMax, TikTok Smart+) have automated bidding, targeting, and placement. Creative variation is the remaining human lever.
• Incrementality Testing — The Causal Layer Under Marketing Attribution Incrementality testing measures the *causal* contribution of a marketing channel — what would happen if you turned the campaign off. Distinct from attribution, which is correlational. The three main test designs are geo-holdout, audience-split, and time-based. When MMM and incrementality disagree, incrementality wins. The 2026 best practice: fewer tests that materially change decisions, not more tests.
• Marketing Mix Modeling — Top-Down Statistical Attribution Marketing Mix Modeling (MMM) estimates the contribution of every marketing channel to revenue using aggregate spend and outcome data — no cookies, no pixels, no user tracking. Adoption surged 212% since 2023 because cookie deprecation broke last-click; MMM doesn't need tracking. Google's Meridian (2024) and Meta's Robyn democratized what was a six-figure consulting engagement.
• Meta Advantage+ — The AI Ad-Automation Suite (Sales, App, Leads + Creative) Meta Advantage+ is the umbrella brand for Meta's AI ad automation: end-to-end campaign types (Sales, App, Leads) plus single-step features. By 2026 it's the enforced default.
• Meta Advantage+ Creative — Automatic Ad Enhancements, GenAI, and the Brand-Control Problem Advantage+ Creative is Meta's single-step creative-automation feature: standard enhancements plus generative AI (background/image generation). On by default — and the brand-safety risk is real.
• Multi-Touch Attribution (MTA): What It Is and Why It's Now a Tactical Tool, Not Ground Truth MTA splits conversion credit across the touchpoints in a customer journey. ATT and cookie loss degraded it — so in 2026 it's a tactical signal, triangulated with MMM and incrementality.
• Reference-Image Conditioning — Show, Don't Tell, the AI Aesthetic Controlling AI image/video aesthetics by feeding reference images (composition, palette, structure, style) instead of writing descriptive prose. Covers what each reference slot controls, the main tools (Midjourney sref/cref/cw, Flux Kontext, Soul guided generation), when reference beats prose and when it doesn't — and the brand-coherence move: feeding the client's own assets so the output looks like them.
• Meta Andromeda — The Ad-Retrieval Engine That Made Creative the Lever Meta Andromeda is the ML ad-retrieval engine (detailed Dec 2024) that selects a few thousand ads from tens of millions of candidates, built to handle Advantage+ creative volume.
• AI Creative Reverse-Engineering: Definition and Method AI creative reverse-engineering deconstructs a winning ad's formula — composition, lighting, palette, copy — into a reusable template for your product.
• AI Crawler — Definition An AI crawler is an automated bot that fetches web content for an AI system — to train a model, build a citation index for AI search, or fetch a page a user asked about. The three types determine your access policy.
• Bytespider — Definition Bytespider is ByteDance's (TikTok's parent) web crawler, widely reported to ignore robots.txt and crawl aggressively. It's the canonical example of why robots.txt alone can't stop a non-compliant AI scraper.
• CCBot (Common Crawl) — Definition CCBot is Common Crawl's web crawler. Common Crawl is a nonprofit that publishes a free, open archive of the web — and that archive is a major training-data source for many AI models. CCBot respects robots.txt.
• GPTBot — Definition GPTBot is OpenAI's web crawler that collects content to train its models. It respects robots.txt, publishes its IP ranges, and is distinct from OAI-SearchBot (search) and ChatGPT-User (user-fetch).
• llms.txt — Definition llms.txt is a proposed plain-text/markdown file that gives AI systems a curated map of your site's most important content. It's advisory — it helps comprehension, not access control.
• Pay-Per-Crawl — Definition Pay-per-crawl is Cloudflare's model that lets sites charge AI crawlers for access using the HTTP 402 'Payment Required' status code and crawler-price headers — turning bot access into a transaction instead of a free-for-all.
• WAF (Web Application Firewall) — Definition A WAF is a firewall that inspects and blocks web requests at the edge before they reach your server. For AI bots it's the enforcement layer robots.txt isn't — it acts, robots.txt only asks.
• Automation Eats Execution — The Cross-Domain Pattern in How AI Reshapes Work A Primores-named framework: across multiple marketing and software domains in 2026, AI tools commoditize the high-volume execution layer first, while strategy, judgment, and integration work stays human-leveraged. The labor-economics implication: execution-tier compensation flattens; strategy-tier compensation rises.
• GEO/AEO (Generative/Answer Engine Optimization) — What It Means Optimizing content for AI search engines like Perplexity, ChatGPT, and Google AI Overviews
• E-E-A-T — Experience, Expertise, Authoritativeness, Trustworthiness E-E-A-T is Google's quality framework — and in 2026 it behaves less like a soft ranking signal and more like a near-binary filter on whether AI engines will cite you at all.
• Retrieval vs Citation — Why Being Fetched by AI Isn't Being Cited AI assistants fetch dozens of pages per query but cite only about half. ChatGPT cited ~50% of retrieved URLs in Ahrefs' 2026 analysis; the other half feed the answer as uncredited background. What predicts which retrieved page gets the citation: title-to-prompt semantic match and readable URL slugs. The gap reframes GEO from 'get retrieved' to 'get cited.'
• The 'Best X' Listicle — The Most-Cited Content Format in AI Search Why 'Best [category]' comparison listicles are the single most-cited content format in AI chatbots (43.8% of ChatGPT's cited page types per Ahrefs 2026), how top-third placement drives recommendation, why freshness is load-bearing, and the strategy: own the list in your category or get placed highly in third-party lists.
• Win-Loss Analysis — The Highest-Leverage CI Layer Win-loss analysis interviews recent prospects (both won and lost deals) to understand the actual reasons behind decisions. The CI layer practitioners credit most with moving win rate — the mechanism (structured retrospective review) is anchored in tier-1 AAR/debrief meta-analyses, though the exact win-rate magnitude is a practitioner figure. Klue/Crayon convergence on 8 best practices: organizational buy-in, defined goals, senior decision-makers as respondents, deals competed to the bitter end, direct buyer interviews (not sales reps), 30–90 day timing, multi-source triangulation, quantitative + qualitative analysis.
• Appropriate Reliance — Trusting AI the Right Amount, Not the Most The goal with AI isn't maximum reliance — it's calibrated reliance. Peer-reviewed evidence shows mere AI labeling triggers costly over-reliance and suppresses critical thinking, yet experts under-rely, and disclosing AI use erodes trust. The reconciliation: appropriate reliance, moderated by expertise and stakes.
• Customer-Perception Moments — How Style, Timing, and Structure Shape Judgment A Primores framework consolidating the wiki's behavioral-evidence research on customer perception: at discrete moments of judgment — the decision moment, the review-writing moment, the failure-recovery moment — small choices about content style, timing, and display structure have outsized, peer-reviewed effects. The cross-cutting meta-pattern: every headline behavioral finding comes with a context-dependent moderator (hedonic-vs-functional, severity, focal-customer) that can flip it. The practitioner discipline is to identify the moment and its moderators before applying the headline.
• Review Response Strategy — How to Reply to Reviews (Backed by ISR Research) Replying to reviews isn't just customer service — it's public theater. Peer-reviewed research (Information Systems Research) shows responses lift future review volume via a third-party effect, should be detailed for negative and brief for positive, and must match tone to the type of unfairness.
• AI Humor and Forgiveness — Self-Deprecating Humor as a Service-Failure Recovery Tactic When an AI agent makes a mistake, humorous responses make users more forgiving — and self-deprecating humor outperforms positive humor by a wide margin. Xie et al. 2025 (n=1,919, Journal of Business Research) found +47.8% forgiveness uplift for self-deprecating vs no-humor on low-severity errors. The effect disappears for high-severity failures and inverts when the customer is the focal victim (Honora et al. 2025, J. Business Ethics) — humor reads as sarcasm and reduces perceived company morality. The 2026 practitioner gate: low severity + non-focal-customer-burned + AFTER resolution = humor helps. Otherwise: stay sincere.
• The Weekend Review Effect — Timing, Anchoring, Incentives, and the 2026 Review-Ops Cluster Online reviews submitted on weekends average 3% lower share of 5-star ratings and 6% higher share of 1-3 star ratings (Bayerl et al. 2026, Journal of Marketing Research, n=400M reviews / 33 platforms). Effect reverses for hedonic products (2023 counter-finding, n=588K) and conflicts with industry response-rate data showing Saturdays among the highest-volume send-days. The wiki's integrated 2026 review-ops cluster: weekend-effect + first-review anchoring + incentive-positivity + display-order — what to do with each, in what order, and when each effect matters practically.
• Continuous Competitive Monitoring — The 2026 CI Discipline Always-on tracking of competitor signals across web, pricing, product, hiring, ads, and reviews — replacing the stale quarterly competitive-landscape deck. The 2026 bottleneck is no longer data collection (Crayon tracks 100+ signal types per competitor); it's signal-to-noise triage and routing intelligence to the people who decide. AI compresses synthesis; decision-linkage discipline stays human.
• Creative Reverse Engineering — Extracting Structural Patterns from Competitor Ads The systematic discipline of deconstructing competitor ad creative to extract reusable structural patterns — composition, lighting, focal hierarchy, copy skeleton — without copying the brand-specific skin. The 2026 vision-LLM stack (Claude + GPT-4o + Foreplay / Atria / Motion + Meta Ad Library) compressed what used to be an art-director consulting engagement into an afternoon. Pattern extraction across many examples is the load-bearing move; single-ad deconstruction underperforms.
• Battlecards — The Living Sales-Enablement Layer Sales battlecards are one-to-two-page references that arm reps with competitive intelligence during live deal conversations. The 2026 evolution: static battlecards are dying; living battlecards are modular, AI-assisted, role-specific, with governance metadata. Updated monthly minimum (weekly for fast markets). Measured by win-rate-against-named-competitor — not by completeness. If a battlecard doesn't fit on one screen, reps won't use it mid-call.
• Prompt Caching — The Production Cost-Optimization Layer for LLM Applications Prompt caching reuses LLM input tokens across requests, cutting input-token costs by up to 90% (Anthropic cache reads are 10% of base price; OpenAI cached inputs run 75-90% cheaper). Combined caching strategies achieve 70-80% total cost reduction in production. The 2026 production landscape: Anthropic cache_control markers with 5-min default TTL (1-hour extended), OpenAI automatic prompt caching, semantic caching via vector similarity (Redis, GPTCache). Distinct from KV caching (model-internal) and agentic memory (cross-session persistence).
• Cohort Analysis — Reading the Shape, Not the Average Cohort analysis groups customers by acquisition date (or other attribute) and tracks them over time. The shape of the retention curve — particularly the M0–M3 onboarding cliff — drives LTV more than any other input. Surviving-cohort LTV (M3+) paired with NRR and CAC payback is a better business-health signal than any single aggregate LTV. 2026 benchmarks: B2B SaaS 3.2:1 LTV:CAC median, DTC subscription 4.1:1 (parity with SaaS reached this year).
• Embeddings — How AI Turns Meaning Into Numbers Embeddings are numerical representations of text (or images, audio, video) that preserve semantic similarity. Two texts about the same topic have similar embedding vectors, even when they use different words. Embeddings are the foundational layer under semantic search, RAG, recommendation systems, and most AI applications that need to match meaning rather than match keywords.
• Guardrails — The Production Safety Layer for AI Systems Guardrails are the technical and operational controls that bound what AI agents can do, when, and with what permissions. They are the production safety layer that sits between a capable model and the real world. Without guardrails, capable agents become uncontrollable. With well-designed guardrails, the same agents are reliable enough to ship. Guardrails are paired with tool use — every powerful tool needs a corresponding guardrail.
• Hallucination — When AI Confidently Invents Things Hallucination is when AI generates content that is plausible-sounding but false. The structural cause is that LLMs predict probable next tokens rather than retrieve true facts. Inside the model's training distribution, hallucination is rare; outside it (uncommon topics, specific named entities, recent events) it's reliable. Hallucination is what makes the jagged-frontier asymmetry dangerous — wrong answers look identical to right ones.
• Tool Use — How AI Agents Reach Out of the Model Tool use is the capability that lets an AI model call external functions — search, calculators, APIs, databases, code execution — rather than generating answers from training alone. Tool use is what turns a chatbot into an agent. It is the technical foundation under all agentic-commerce, agent-engineering, and managed-agents work. The discipline of designing tools well determines whether an agent is reliable or theatrical.
• Agent Adoption Frictions — Three Psychological Barriers Wharton's 2026 framework: AI agent adoption is blocked by three psychological frictions — perceived competence, trust, and delegation of control. The barrier is not the technology; it's whether users will hand over the keys.
• Agent Engineering — Karpathy's Ceiling-Raising Discipline Andrej Karpathy's framing (Sequoia AI Ascent 2026) for the professional discipline of coordinating AI agents reliably and safely — distinct from vibe-coding. Vibe-coding raises the floor of who can build software; agent engineering raises the ceiling of what professionals can deliver.
• Vibe Coding — Building Software by Describing What You Want Andrej Karpathy-coined term (Feb 2025) for AI-assisted coding where you describe intent and accept the AI's output without deeply reading the code. Karpathy's May 2026 update (Sequoia AI Ascent): vibe coding raises the floor of who can build software; the production-side discipline is agent engineering.
• Advisor Strategy — Pairing a Smarter Model as an Occasional Advisor With a Cheaper Executor Anthropic's advisor pattern (April 2026): the executor model (Sonnet or Haiku) handles tasks end-to-end while consulting an advisor model (Opus) only on hard decisions. Server-side, single API request. Sonnet+Opus advisor: +2.7pp on SWE-bench at -11.9% cost. Haiku+Opus: 41.2% on BrowseComp vs 19.7% solo, 85% cheaper than Sonnet alone.
• AI Skill Leveling — Why Novices Gain Most From AI Tools Across three independent studies (Brynjolfsson 2023 n=5,179, Noy-Zhang 2023 n=444, Dell'Acqua 2023 n=758), AI tools systematically lift novice/low-performer productivity more than expert productivity. The skill premium compresses. The implications for hiring, training, and agency pricing are direct.
• Focal Hierarchy in Ad Design: What It Is and How It Works Focal hierarchy is the ordering of visual elements so the viewer's eye lands on them in sequence — product first, supporting cue second, CTA third.
• Jagged Frontier — Why AI Helps on Some Tasks and Hurts on Others The 'jagged technological frontier' (Dell'Acqua et al. 2023) is the empirical finding that AI improves performance on tasks inside its capability boundary but degrades performance on tasks just outside it — and the boundary is invisible from the outside. Direct evidence for the automation-eats-execution thesis.
• Lighting Recipe in Ad Photography: Definition and Why It Matters A lighting recipe specifies key direction, fill, rim, color temperature, and contrast. It's the highest-leverage transferable element in ad creative.
• Product Framing Archetypes: The 8 Patterns Every Marketer Knows A framing archetype is a reusable way of staging a product in an ad — hero, lifestyle, macro, levitation, flatlay, hand-held, before/after, founder-selfie.
• Recognition-Primed Decision Making — When Expert Intuition Is Reliable Klein's RPD model (1989/1998) describes how experts make rapid decisions by pattern-matching against past experience, not by comparing options. The Klein-Kahneman 2009 synthesis specifies the conditions under which intuition is reliable: high-validity environments + prolonged practice with rapid feedback. Predicts where AI itself will struggle for the same reasons.
• Distinctive Assets — What They Are and Why They Beat Differentiation Brand-specific cues (colors, logos, fonts, tone, mascots, jingles) that trigger recognition without active thought. Sharp's argument: distinctiveness builds mental availability; differentiation usually doesn't matter empirically.
• Double Jeopardy Law — Why Smaller Brands Get Hit Twice Smaller brands have fewer buyers AND those buyers buy slightly less often. Loyalty doesn't vary much across competing brands. Penetration is the lever for growth, not loyalty.
• Dual-Process Thinking — System 1, System 2, and Why Scroll Behavior Is Pure System 1 Kahneman's framework: System 1 is automatic and fast (intuition, perception, heuristics); System 2 is effortful and slow (deliberation, calculation). Most content decisions happen entirely in System 1. The wiki's existing concepts (mental availability, scarcity, information scent) all rest on this substrate.
• Information Foraging — What It Means People seek information the way predators seek food: optimizing rate of valuable information gained per unit cost. Explains scroll behavior, diet selection, and why volume alone doesn't drive attention.
• Mental Availability — What It Is and Why It Drives Brand Growth A brand's propensity to be thought of in buying situations. Sharp's central thesis: brands grow by maximizing mental availability across many light buyers, not by deepening loyalty in a small core.
• Persuasion Principles — The Six Cialdini Levers Of Compliance Six psychological principles (reciprocation, commitment & consistency, social proof, liking, authority, scarcity) that trigger near-automatic compliance. Cialdini's framework applies wherever a specific persuasion moment matters — clicks, shares, sales, conversions.
• TPB (Theory of Planned Behaviour) — What It Means Plain-English explanation of how attitudes, social norms, and perceived control drive consumer decisions
• Weak Ties — What They Are and Why They Diffuse Information Granovetter's claim that bridges between social clusters are necessarily weak ties. Information diffuses through acquaintances, not close friends. Algorithmic feeds operate as synthetic weak-tie bridges.
• Super-Niche — What It Means A content niche defined as Audience × Problem × Context — narrow enough to dominate, broad enough to sustain
• Topical Authority — What It Means SEO strategy of publishing exhaustive interlinked content on one topic to become Google's recognized expert
• Astroturfing — Fake Grassroots Marketing Coordinated promotion disguised as organic community engagement. How to detect it and why authentic marketing wins long-term.
• Substance Ranking — Content Quality Over Popularity A content evaluation method that scores comments and posts on evidence quality rather than engagement metrics like upvotes or likes
• GEO Anchor — First-Sentence Citation Optimization A content pattern that makes AI search engines more likely to cite your content: answer directly in sentence one
• Honest Assessment — AI Trust Signal Why admitting product weaknesses increases AI search citations: the counter-intuitive GEO pattern
• Rumpelstiltskin Effect — Why Naming Problems Drives Sales The psychological principle that giving a customer's problem a specific name builds trust, reduces anxiety, and positions your brand as the solution
• SMRA (Social Media Recommendation Algorithms) — What It Means Plain-English explanation of how social media algorithms personalize content and their psychological effects
• Cognitive Automation — What It Means Plain-English explanation of cognitive automation for business professionals
• Fine-Tuning — What It Means Plain-English explanation of LLM fine-tuning for business professionals
• LLM Evals — Evaluation Systems for AI Products Plain-English guide to building evaluation systems that make AI products actually work
• LLM Nudges — How AI Guides User Decisions Understanding the follow-up suggestions AI systems use to shape user behavior and customer journeys
• RAG (Retrieval-Augmented Generation) — What It Means Plain-English explanation of RAG for business professionals
• Skill — Reusable AI Instruction Package A folder containing instructions that teach Claude specific workflows, enabling consistent AI behavior across conversations
• Tokens — What They Mean Plain-English explanation of LLM tokens for business professionals
• Agent Outcomes — What It Means A pattern where AI agents work toward defined completion criteria, with a separate grader evaluating success
• AI Agent — What It Means Plain-English explanation of AI agents for business professionals
• LLM (Large Language Model) — What It Means Plain-English explanation of Large Language Models for business professionals
• LLM Wiki Pattern — What It Means A pattern for building compounding knowledge bases using LLMs
• Prompt Engineering — What It Means Plain-English explanation of prompt engineering for business professionals
• Zettelkasten — What It Means A note-taking methodology of interconnected atomic notes, perfect for LLM-maintained knowledge bases