AI Purchase Channel — Key Terms

The practice of optimising content so that AI assistants and search engines surface it when directly answering user questions. AEO builds on traditional SEO but shifts the goal from ranking in results pages to being cited or quoted within AI-generated answers. Related to GEO but typically narrower in scope — AEO focuses on factual, definitional, and how-to questions, while GEO also covers purchase journey queries.

An AI system that takes autonomous actions in the world — not just generating text in response to a prompt, but executing multi-step tasks: browsing the web, interacting with services, completing purchases, and operating persistently without a human in the loop for each step. Agentic AI is the direction consumer AI assistants are moving: from ChatGPT answering questions to AI completing shopping tasks on a user's behalf.

A web crawler operated by an AI company to collect content for training data or real-time retrieval. Major AI crawlers include GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot (Perplexity), and GoogleBot-Extended (Google Gemini). Pages that block these crawlers via robots.txt may not be cited by those AI providers.

The AI-generated summaries that appear at the top of some Google search results pages, produced by Google's Gemini model. AI Overviews synthesise information from multiple sources into a direct answer, with cited sources linked below. Appearing as a cited source in AI Overviews is a distinct optimisation goal from traditional SEO ranking.

The new buying channel where shoppers — human or AI agent acting on a human's behalf — ask AI assistants (ChatGPT, Perplexity, Google Gemini, Google AI Mode, Amazon Rufus, and others) what to buy. The channel is conversational rather than query-based, opinionated by default (one recommendation, not ten links), and personalised through assistant memory. Crucially, the channel user is no longer always a person: Operator, Computer Use, ChatGPT tasks, and other agents now research and transact on shoppers' behalf. The brand-side problem is identical for both — be the recommendation. Slingso exists specifically to help brands win this channel.

Read the full definition →

A broad term for search experiences powered by large language models, where results are synthesised responses rather than ranked lists of links. Includes Google AI Overviews, Perplexity, ChatGPT Browse, and Google AI Mode. Distinct from traditional search (10 blue links) in that the AI interprets, synthesises, and presents information rather than listing sources.

A measure of how frequently a brand, product, or concept appears in AI-generated responses. Typically expressed as a mention rate or share of voice across a defined set of queries and AI providers. AI visibility is a useful directional metric but is a leading indicator of AI channel performance — the downstream metric that matters is purchase likelihood and attributable revenue.

The mandatory brand review step in Slingso's Create loop. After the agent generates content, it stops and places every piece in the brand's approval queue. Nothing is published until the brand explicitly approves it. The approval gate is an architectural constraint, not a settings toggle — it cannot be bypassed.

A piece of content created by Slingso's Create loop and placed in the brand's approval queue. Artifacts include long-form articles, product FAQ pages, comparison content, updated product descriptions, and PR briefing notes. Each artifact is linked to the specific scenarios and purchase likelihood gaps it was created to address.

The accumulated, structured knowledge the Slingso agent builds about a brand across every agent run. Brand memory includes: product positioning and attributes, approved tone and voice patterns, competitor intelligence, and the historical impact of past interventions on purchase likelihood. Brand memory compounds — each cycle makes the agent more effective than the last.

An AI agent that acts on behalf of a consumer to research, compare, and complete purchases. Buyer agents represent the next evolution of the AI purchase channel: instead of a human asking an AI for recommendations and then visiting a website to buy, the AI agent conducts the entire purchase journey autonomously. OpenAI Operator, Perplexity's shopping features, and Amazon Rufus are early implementations.

An instance of an AI assistant referencing a specific source — a URL, publication, brand, or product — within an AI-generated response. Citations are the primary mechanism through which AI models attribute information and recommendations to specific sources. Being cited is a prerequisite for AI-driven product discovery.

The specific URL or domain that an AI assistant cites as the basis for information in its response. When ChatGPT recommends a product and cites your product page or an article about your brand, those URLs are citation sources. Identifying which citation sources AI models use — and whether your brand controls any of them — is a core function of AI visibility tools.

The output of Slingso's Analyse loop: a prioritised document explaining exactly what content to create, which specific scenarios it targets, what competitors are doing that the AI is rewarding, and what signals (product features, certifications, use cases, pricing, availability) the content must include. The Create loop takes the content brief as its primary input.

A specific query, topic, or question for which a brand has no credible content that AI models can cite or reference. Content gaps are the root cause of low purchase likelihood — when an AI assistant looks for content to inform a product recommendation and finds nothing from a specific brand, it defaults to citing competitors who have covered the topic.

The practice of creating and structuring content to be cited and recommended by generative AI systems — ChatGPT, Gemini, Perplexity, Claude, and others. GEO extends traditional SEO by optimising for AI-generated answers rather than keyword ranking. Key GEO signals include: directness and specificity of answers, inclusion of original data or unique claims, authoritative citations, structured markup, and conversational format.

The process of connecting an AI model's response to specific, verifiable source documents rather than relying solely on the model's training data. Grounding reduces hallucination and increases citation accuracy. For brands, grounding is important because AI models that ground their responses in web content will cite your brand more accurately — and more frequently — if your published content contains clear, factual, well-structured claims.

An instance where a large language model generates information that is factually incorrect, invented, or unsupported — presented with the same apparent confidence as accurate information. Hallucination is a known limitation of LLMs. For brands, it has two implications: (1) AI assistants may hallucinate positive attributes for competitors that do not exist, and (2) AI assistants may hallucinate incorrect product information for your brand if your published content is absent or ambiguous.

A brand's primary revenue-driving SKU or the product with the highest strategic importance for AI Purchase Channel performance. Slingso tracks a defined number of hero products per plan — 2 on Starter, 6 on Growth, 12 on Scale. Hero products are the ones with the most to gain from being recommended in the AI Purchase Channel because they appear in the highest-volume purchase-intent queries.

A neural network trained on large volumes of text data to generate, summarise, translate, and reason about language. LLMs power all major AI assistants — GPT-4 (ChatGPT), Claude (Anthropic), Gemini (Google), and Llama (Meta). When a consumer asks an AI assistant what product to buy, the LLM processes the query, retrieves relevant context, and generates a response — which may or may not include your brand.

The practice of improving how a brand is represented within large language model responses — not just in web-indexed content, but across the sources, formats, and signals that LLMs weight when generating recommendations. Slingso's primary agent is the LLM Optimization Agent: it optimises your brand's purchase likelihood across the major consumer LLMs.

The first of Slingso's four agent loops, running daily. The Monitor loop submits all configured scenarios to each configured AI provider, records the full AI response, extracts which products were recommended and in what context, and calculates purchase likelihood for every tracked product on every scenario. The Monitor loop feeds observations to the Analyse loop.

A piece of content created by the Slingso agent and placed in the brand's approval queue. Outputs per month are capped by plan: 10 on Starter, 30 on Growth, unlimited on Scale. Each output is tied to a specific content brief and a specific set of scenarios it is intended to improve.

A defined representation of a target customer used by the Slingso agent when simulating purchase journeys. Each persona includes: demographic profile (age, location, occupation), purchase goals and concerns, price sensitivity, typical research behaviour, and preferred channels. Personas are derived from a brand's site during onboarding and refined over time. Scenarios are always run from the perspective of a specific persona.

An AI assistant platform against which Slingso runs scenarios. Current providers on public plans: ChatGPT, Google AI Overviews, Google Gemini, Perplexity, and Grok. Providers are added by plan tier — Starter covers ChatGPT + Google AI Overviews + Gemini; Growth adds Perplexity; Scale adds Grok. Additional assistants (Claude, MS Copilot, Google AI Mode, DeepSeek, Meta AI) are available on custom/enterprise plans. Different providers give different recommendations for the same query — monitoring each separately is essential.

A query submitted to an AI assistant that explicitly or implicitly signals a purchase decision. Examples: 'best creatine monohydrate for beginners', 'what protein powder should I buy for weight loss', 'compare Brand X and Brand Y vitamin C serum'. Purchase intent queries are the highest-value scenarios to monitor and optimise because they directly influence the purchase outcome.

The process of simulating a full consumer AI shopping journey from a defined persona's perspective — from initial awareness through research, comparison, and purchase intent — across one or more AI providers. Slingso's Monitor loop runs purchase journey simulations daily for each configured scenario. Unlike single-query monitoring, journey simulation captures how AI recommendations evolve across a multi-turn conversation.

The primary performance metric in Slingso: a score from 0–100 measuring how likely a specific AI assistant is to recommend your product as the primary purchase for a given scenario. It is calculated from appearance rate, position in the recommendation, recommendation strength language, and whether a purchase action was included. The target is maximising purchase likelihood on your highest-value scenarios across your most important providers.

A framework for AI agents that interleaves reasoning (thinking through what to do next) with acting (actually doing it — calling tools, scraping pages, writing content). Slingso's Analyse and Create loops run on a ReAct architecture: the agent reasons about what information it needs, takes an action to get it, observes the result, reasons again, and iterates — up to a defined turn limit.

The core unit of measurement in Slingso: a specific combination of query × GEO (geographic market) × target persona (optional) × target product (optional). Scenarios do not include AI providers — providers are a separate dimension that multiplies how many times each scenario runs each day. Starter: 40 scenarios × 3 providers = 120 daily runs. Growth: 120 × 4 = 480 daily runs. Scale: 200 × 5 = 1,000 daily runs. A scenario is what other AI-visibility tools call a 'prompt' or 'query'.

The structured, distilled knowledge the Slingso agent stores across all Monitor and Analyse runs — product-level summaries of what the AI currently knows and says about your brand, what content is being cited, and which competitive signals are driving gaps. Semantic memory is updated after every monitor cycle and used by the Analyse and Create loops to prioritise actions.

The percentage of AI responses — across a defined set of queries and providers — in which a brand's product is mentioned or recommended, expressed relative to competitor brands. AI share of voice is a visibility metric. It tells you how often you appear relative to competitors but does not directly measure purchase outcomes. Slingso tracks share of voice as context alongside the primary metric of purchase likelihood.

Monitoring and optimisation conducted at the individual product level (Stock Keeping Unit) rather than at the brand level. SKU-level tracking is important because AI assistants recommend specific products, not brands in general. A brand's flagship product may have a purchase likelihood of 80 while a newer product in the same category has a likelihood of 10 — and fixing that gap requires product-specific analysis and content.

A consumer discovery or purchase interaction that occurs entirely within an AI assistant's interface, without the consumer ever visiting the brand's website. When an AI assistant recommends a product, provides sufficient detail about it, and includes a direct purchase link, the entire purchase journey may complete without a single brand website visit. Zero-click discovery makes traditional web analytics an incomplete measure of AI channel performance — revenue may be attributable to AI recommendations that never generated a session in Google Analytics.