Glossary of conversational AI

• Cart abandonment
• AI agent
• Revenue attribution
• Conversational commerce
• CSAT (Customer Satisfaction Score)
• Human handoff (escalation)
• Customer experience (CX)
• Dynamic FAQ
• Fine-tuning
• Hallucination (AI)
• Conversational AI
• Sovereign AI
• Unified Inbox
• LLM (large language model)
• NPS (Net Promoter Score)
• Average order value (AOV)
• Prompt
• RAG (Retrieval-Augmented Generation)
• GDPR and AI
• Session (session-based billing)
• Trust signals
• Omnichannel support
• E-commerce conversion rate
• Deflection rate
• First contact resolution rate (FCR)
• First response time (FRT)
• Ticket (customer support)
• Chat widget

I

Cart abandonment

Cart abandonment is when a customer adds items to their online shopping cart but leaves the website before completing the transaction. It’s a common behavior and can have various causes: unexpected shipping costs, an overly complex checkout process, distractions, or simply looking for information with no immediate intent to buy. It’s a key indicator of friction in the customer journey and a major source of lost potential revenue for online retailers. Understanding and reducing cart abandonment is essential to optimizing an online store’s performance.

Why it matters. Cart abandonment means missed sales, directly. By identifying friction points and proactively engaging customers through AI agents, you can recover a significant share of that lost revenue, improve the shopping experience and increase your overall conversion rate.

See also: E-commerce conversion rate, Average order value (AOV), Conversational commerce

II

AI agent

An AI agent is an autonomous software program designed to perform specific tasks or functions, often interacting with users or other systems. It uses artificial intelligence, including machine learning and natural language processing, to understand requests, process information and carry out actions. These agents can operate in a variety of environments, from customer service chatbots to complex automation systems, learning and adapting over time to improve their performance and efficiency. Their goal is to automate repetitive processes, provide instant assistance and improve the user experience by simulating human interactions and decision-making abilities.

Why it matters. An AI agent can handle a high volume of customer requests 24/7, reducing operating costs and improving satisfaction. It frees your teams for more complex tasks, while providing instant, personalized assistance, which is essential for loyalty and conversion in e-commerce.

See also: Conversational AI, Human handoff (escalation), Customer experience (CX)

III

Revenue attribution

Revenue attribution is the process of identifying and assigning the value of a sale to one or more marketing touchpoints that contributed to the customer’s purchase decision. It lets businesses understand how effective their various marketing channels and campaigns are. There are several attribution models, such as first touch, last touch, or multi-touch models that spread credit across the entire customer journey. Accurate attribution is crucial for optimizing marketing budgets, improving return on investment (ROI) and refining acquisition and retention strategies.

Why it matters. Understanding revenue attribution is vital to optimizing your marketing spend and maximizing your ROI. It lets you know which channels and interactions actually generate sales, and adjust your strategies for more efficient growth. AI agents can provide valuable data on the customer interactions that lead to conversion.

See also: E-commerce conversion rate, Customer experience (CX), AI agent

IV

Conversational commerce

Conversational commerce is an e-commerce strategy that builds dialogue-based interactions (through chatbots, voice assistants or human agents) into the entire customer journey. It lets consumers discover products, ask questions, receive personalized recommendations and even complete purchases using natural language, often within messaging apps or chat widgets. The goal is to create a smoother, more personalized and more engaging shopping experience, mimicking human interaction to ease the purchase decision and customer support.

Why it matters. Conversational commerce transforms the shopping experience by making it more intuitive and personalized. By letting customers interact naturally with an AI agent, you can increase engagement, provide immediate and relevant support, and drive sales by effectively guiding users toward their purchases.

See also: AI agent, Conversational AI, Chat widget

V

CSAT (Customer Satisfaction Score)

CSAT, or Customer Satisfaction Score, is a key performance indicator that measures a customer’s level of satisfaction following a specific interaction, product or service. It’s usually collected through a direct question such as “Are you satisfied with your experience?” with a response scale (for example, from “very dissatisfied” to “very satisfied,” or a numeric scale). The score is calculated as the percentage of positive responses. CSAT is a barometer of short-term satisfaction, making it possible to quickly identify strengths and areas for improvement in the customer journey.

Why it matters. For your e-commerce business, a good CSAT is essential because it reflects the perceived quality of your services and products. An AI agent can improve this score by offering fast, relevant answers, resolving requests efficiently and personalizing the customer experience, turning interactions into positive moments.

See also: Customer experience (CX), NPS (Net Promoter Score), Deflection rate

VI

Human handoff (escalation)

Human handoff, or “escalation,” is the process by which a customer interaction, initially handled by an AI agent or a chatbot, is transferred to a human advisor. This transfer happens when the AI agent can’t resolve the request because of its complexity, its specificity, or the need for emotional intervention or human expertise. The goal is to ensure a smooth transition for the customer, making sure their request is ultimately handled satisfactorily, with no break in the support experience.

Why it matters. Human handoff is essential to ensure optimal customer satisfaction, even with an AI agent. It lets you handle complex or sensitive cases, avoiding frustration and building trust. A smooth transition to a human ensures every customer gets the help they need, protecting your e-commerce business’s reputation.

See also: AI agent, Customer experience (CX), First contact resolution rate (FCR)

VII

Customer experience (CX)

Customer experience (CX) encompasses all the perceptions and feelings a customer develops during all their interactions with a brand, from initial discovery to after-sales service. It includes every touchpoint, physical or digital, and directly influences satisfaction, loyalty and the likelihood to recommend. A positive CX is characterized by ease, efficiency, personalization and emotion, and has become a key differentiator in a competitive market, directly affecting a company’s business performance.

Why it matters. An excellent customer experience drives loyalty and growth for your e-commerce business. By offering smooth, personalized and efficient interactions, especially with an AI agent, you turn visitors into repeat buyers and brand ambassadors, increasing your revenue.

See also: AI agent, CSAT (Customer Satisfaction Score), NPS (Net Promoter Score)

VIII

Dynamic FAQ

A dynamic FAQ is an evolved version of the traditional Frequently Asked Questions page, using artificial intelligence and natural language processing (NLP) to provide personalized, contextual answers to users’ questions. Unlike a static FAQ, it doesn’t just display a predefined list of questions and answers: it interprets natural-language queries, searches for relevant information in an extensive knowledge base and generates precise answers in real time. It learns and improves continuously through interactions, offering a more intuitive and effective self-service experience.

Why it matters. A dynamic FAQ is a major asset for your e-commerce business, because it lets customers find answers to their questions instantly, reducing the need to contact support. Powered by an AI agent, it improves user autonomy and relieves your teams of repetitive requests, optimizing the customer experience.

See also: AI agent, Conversational AI, RAG (Retrieval-Augmented Generation)

IX

Fine-tuning

Fine-tuning is a machine learning technique that adapts a pre-trained AI model (such as a large language model) to a specific task or domain by training it further on a smaller, targeted dataset. Rather than building a model from scratch, fine-tuning lets you leverage the general knowledge acquired by the initial model, then specialize it for precise needs, improving its performance and relevance for particular applications without requiring massive training resources.

Why it matters. Fine-tuning customizes your AI agent so it speaks your brand’s language and understands the specifics of your products and customers. It ensures more accurate, relevant answers, improving support efficiency and the customer experience, which can boost sales and loyalty.

See also: LLM (large language model), RAG (Retrieval-Augmented Generation), Prompt

X

Hallucination (AI)

In artificial intelligence, a “hallucination” is the tendency of a model, especially a large language model (LLM), to generate information that seems plausible but is factually incorrect, made up, or disconnected from reality. These errors are not intentional; they result from the model’s limits in telling truth from fiction or in accessing precise data. Hallucinations can undermine the reliability of AI systems, particularly in applications that demand high factual accuracy, such as customer support or information retrieval.

Why it matters. Hallucinations can seriously damage your customers’ trust and your brand’s reputation by delivering wrong information. It is crucial to use AI agents designed to minimize this risk, for example through RAG, to ensure reliable answers and maintain a positive customer experience.

See also: RAG (Retrieval-Augmented Generation), LLM (large language model), Customer experience (CX)

XI

Conversational AI

Conversational AI is a technology that lets computers understand, process, and respond to human language in a natural way. It combines natural language processing (NLP), natural language understanding (NLU), and natural language generation (NLG) to simulate human conversations. Used in chatbots, virtual assistants, and voice interfaces, it aims to deliver a smooth, intuitive interaction that can solve problems, answer questions, or carry out tasks by interpreting the user’s context and intent.

Why it matters. Conversational AI improves the customer experience by offering instant, personalized interactions, available 24/7. It guides customers, answers their questions, and makes buying easier, increasing engagement and potentially your online store’s conversion rate.

See also: AI agent, LLM (large language model), Prompt

XII

Sovereign AI

Sovereign AI refers to artificial intelligence systems whose development, hosting, and operation are fully controlled and managed within a specific national or regional jurisdiction. The goal is to ensure that the data processed, the algorithms used, and the decisions made by the AI comply with local laws, ethical standards, and data security requirements, particularly regarding privacy and personal data protection. This lets a country or region keep its strategic and technological autonomy, avoiding dependence on foreign powers for critical infrastructure.

Why it matters. Adopting sovereign AI is crucial for the security of your data and for regulatory compliance, notably with the GDPR. It strengthens your customers’ trust by guaranteeing that their information is processed in France, under French jurisdiction, and protects you from the risks of foreign legislation.

See also: GDPR and AI, AI agent, LLM (large language model)

XIII

Unified Inbox

A unified inbox is a centralized interface that gathers all customer communications from different channels (email, live chat, social media, phone, etc.) in one place. Its goal is to give support agents a complete, consistent view of every interaction with a given customer, whatever the channel used. This prevents loss of context, improves agent efficiency, and delivers a smoother, more personalized customer experience, by ensuring every request is handled with all the relevant information at hand.

Why it matters. For your e-commerce business, a unified inbox is essential to efficiently handle a growing volume of requests across various channels. It lets your support teams be more productive and deliver a consistent customer experience, with a full history of exchanges and no juggling between platforms.

See also: Omnichannel support, Ticket (Customer Support), Customer experience (CX)

XIV

LLM (large language model)

A Large Language Model (LLM) is a type of artificial intelligence model trained on enormous amounts of text data to understand, generate, and manipulate human language. These models can perform a wide range of language tasks, such as translation, writing, answering questions, or summarizing information. Their complex architecture lets them identify patterns and semantic relationships, making their interactions and text outputs remarkably coherent and contextual.

Why it matters. LLMs are the brain of your AI agent, enabling smooth, intelligent customer interactions. They understand complex requests and generate relevant answers, improving support efficiency and the quality of the customer experience, which can directly impact your sales.

See also: Conversational AI, RAG (Retrieval-Augmented Generation), Prompt

XV

NPS (Net Promoter Score)

NPS, or Net Promoter Score, is a customer loyalty metric that measures how likely a customer is to recommend a company, product, or service to those around them. It is measured with a single question: “On a scale of 0 to 10, how likely are you to recommend [Company/product/service name] to a friend or colleague?” Customers are classified as promoters (9-10), passives (7-8), or detractors (0-6). NPS is calculated by subtracting the percentage of detractors from the percentage of promoters. It gives a view of overall satisfaction and organic growth potential.

Why it matters. A high NPS is crucial for your e-commerce business, as it signals strong customer loyalty and positive word of mouth, both drivers of growth. An AI agent can help improve it by delivering a smooth, personalized customer experience, turning customers into ambassadors for your brand.

See also: Customer experience (CX), CSAT (Customer Satisfaction Score), E-commerce conversion rate

XVI

Average Order Value (AOV)

Average Order Value (AOV) is a key e-commerce performance indicator that represents the average value of each order placed on a site. It is calculated by dividing total revenue by the total number of orders over a given period. AOV is crucial for assessing sales performance and the effectiveness of sales strategies such as upselling and cross-selling. Increasing average order value grows revenue without necessarily increasing the number of transactions or the traffic on the site.

Why it matters. Increasing average order value is essential to maximize your e-commerce profitability without raising acquisition costs. An AI agent can intelligently suggest complementary products or upgrade options (upselling/cross-selling) to customers during their buying journey, encouraging them to spend more and increasing the value of each order.

See also: E-commerce conversion rate, Conversational commerce, Session (session-based billing)

XVII

Prompt

A prompt is an instruction or question given to an artificial intelligence model, in particular a large language model (LLM), to guide its text generation or response. It can be a single sentence, a detailed paragraph, or even a set of structured data. The quality and clarity of the prompt are crucial, as they directly influence the relevance, accuracy, and format of the output the AI generates. The art of designing effective prompts is called “prompt engineering.”

Why it matters. Well-designed prompts are the key to getting accurate, useful answers from your AI agent, ensuring a smooth customer experience. They steer the AI toward the relevant information and the desired tone, optimizing support efficiency and your customers’ satisfaction.

See also: LLM (large language model), Conversational AI, Fine-tuning

XVIII

RAG (Retrieval-Augmented Generation)

RAG (Retrieval-Augmented Generation) is an AI architecture that combines the power of large language models (LLMs) with the ability to retrieve relevant information from an external knowledge base. Before generating an answer, the RAG system searches for factual data in documents or databases, then uses that information to enrich the LLM’s context. This approach produces more accurate, relevant, and up-to-date answers, while reducing the risk of hallucinations and providing verifiable sources.

Why it matters. RAG lets your AI agent give highly accurate answers based on your own data (product catalog, FAQ, policies). This ensures reliable, consistent information for your customers, improving their trust and reducing errors, which is crucial for satisfaction and loyalty.

See also: LLM (large language model), Hallucination (AI), Dynamic FAQ

XIX

GDPR and AI

The General Data Protection Regulation (GDPR) imposes strict requirements on the processing of personal data, including data collected and processed by artificial intelligence systems. For AI, this means principles of data minimization, algorithmic transparency, informed consent, the right to be forgotten, and data portability. The GDPR also requires a data protection impact assessment (DPIA) for high-risk processing, and regulates automated decision-making, notably profiling, by guaranteeing a right to human intervention and to an explanation. Compliance is essential to avoid heavy penalties.

Why it matters. GDPR compliance is non-negotiable for any company using AI, especially with customer data. Following these rules protects your e-commerce business from steep fines, strengthens your customers’ trust, and ensures an ethical, responsible use of AI, essential for a solid reputation.

See also: Sovereign AI, AI agent, Customer experience (CX)

XX

Session (session-based billing)

In the context of session-based billing, a session is a period of continuous interaction between a user and a service or application, often a conversational agent. It is defined by specific criteria, such as a maximum period of inactivity (for example, 30 minutes without a message) or a limited number of exchanges. Once these criteria are exceeded, the session ends and a new session starts if the interaction resumes. This billing model is commonly used for usage-based services, allowing transparent pricing proportional to the user’s actual activity with the AI agent.

Why it matters. Session-based billing offers great flexibility and cost predictability when using AI agents. You pay only for your customers’ actual engagement, which optimizes your budget and ensures efficient use of resources, with no hidden fees for idle interactions.

See also: AI agent, Conversational AI, Chat widget

XXI

Trust signals

Trust signals are visual or text elements on an e-commerce site, designed to build visitor trust and remove barriers to purchase. They aim to reduce uncertainty and perceived risk, such as payment security, product quality, or delivery reliability. These signals can include secure payment logos, positive customer reviews, guarantees (satisfaction or your money back), clear return policies, quality certifications, or free shipping information. Their presence is crucial to turning visitors into loyal customers.

Why it matters. Trust signals are essential to build confidence and reduce hesitation at purchase. By displaying them clearly, you lower the cart abandonment rate and increase your conversions. An AI agent can also reinforce this trust by providing accurate, reassuring information about your policies and services.

See also: E-commerce conversion rate, Cart abandonment, Customer experience (CX)

XXII

Omnichannel support

Omnichannel support is a customer service strategy that integrates all touchpoints (chat, email, phone, social media, etc.) to deliver a smooth, consistent customer experience, whatever the channel used. Unlike multichannel support, omnichannel keeps the context of the customer interaction from one channel to the next, letting agents pick a conversation up where it was left off, without the customer having to repeat their problem. This customer-centric approach improves satisfaction, loyalty, and operational efficiency by providing a unified view of the customer journey.

Why it matters. Omnichannel support is essential for your e-commerce business to meet the expectations of modern customers who use a variety of channels. It ensures a frictionless experience, where an AI agent can start a conversation on one channel and hand it off with the full context to a human on another, guaranteeing perfect continuity.

See also: Unified Inbox, Customer experience (CX), AI agent

XXIII

E-commerce conversion rate

The e-commerce conversion rate is a key performance indicator (KPI) that measures the percentage of website visitors who take a desired action, usually a purchase. It is calculated by dividing the number of conversions (for example, orders) by the total number of visitors over a given period, then multiplying the result by 100 to get a percentage. This rate is fundamental to assessing the effectiveness of an e-commerce site and its marketing strategies. A high conversion rate indicates a good fit between the offer, the user experience, and customer expectations.

Why it matters. A high conversion rate drives your e-commerce profitability, turning visitors into paying customers. An AI agent can proactively guide users, answer their questions in real time, remove barriers to purchase, and recommend relevant products, significantly increasing the chances of conversion.

See also: Average Order Value (AOV), Cart abandonment, Conversational commerce

XXIV

Deflection rate

The deflection rate measures the proportion of customers who, after trying to contact customer support through a traditional channel (phone, email), are redirected to or find a solution through a self-service channel (FAQ, chatbot, knowledge base). A high deflection rate indicates that customers are able to solve their problems on their own, reducing the workload of human agents and operating costs. It is a key indicator of the effectiveness of self-service tools and of a company’s ability to optimize its customer service.

Why it matters. A high deflection rate means fewer tickets for your teams, reducing operating costs and letting your agents focus on complex cases. An AI agent can considerably increase this rate by instantly resolving common requests, improving efficiency and customer satisfaction on your e-commerce site.

See also: AI agent, Dynamic FAQ, First Contact Resolution (FCR) rate

XXV

First Contact Resolution (FCR) rate

First Contact Resolution (FCR) rate is a performance indicator that measures the percentage of customer requests resolved at the first interaction, without requiring follow-up or additional contacts. A high FCR is a sign of efficient customer service and greater customer satisfaction, as the customer gets a quick solution to their problem. It also reduces operating costs by lowering the number of reopened tickets and repeated interactions. FCR is a pillar of a smooth, frictionless customer experience.

Why it matters. A high FCR is crucial for your e-commerce business, as it maximizes customer satisfaction and reduces your team’s workload. An AI agent can resolve many common requests autonomously and accurately at first contact, freeing your teams for more complex cases and improving overall efficiency.

See also: AI agent, Human handoff, Customer experience (CX)

XXVI

First Response Time (FRT)

First Response Time (FRT) is a key customer service performance indicator that measures the average time between a customer submitting a request and receiving the first response from support. A short FRT is crucial for customer satisfaction, as it reduces waiting and signals to the customer that their request is being handled. It is calculated by dividing the total first-response time by the number of requests. An optimized FRT contributes to a better customer experience and can prevent dissatisfaction from escalating.

Why it matters. For your e-commerce business, a fast FRT is vital to hold customers’ attention and prevent drop-off. An AI agent can provide an instant first response 24/7, informing the customer, solving simple issues, or qualifying the request, which considerably improves the initial experience.

See also: AI agent, Conversational AI, Customer experience (CX)

XXVII

Ticket (Customer Support)

In the context of customer support, a ticket is a formal record of a request, problem, or question submitted by a customer. It is used to track and manage each customer interaction from opening to resolution. Each ticket contains essential information such as the customer’s identity, the nature of the request, the communication history, and the actions taken. Ticketing systems organize the workflow, assign tasks to the right agents, and ensure rigorous follow-up, guaranteeing that no request is forgotten and that issues are resolved in a structured way.

Why it matters. Ticket-based management is fundamental for your e-commerce business to organize and prioritize customer requests, ensuring effective follow-up. An AI agent can create, pre-qualify, or even resolve tickets automatically, reducing your teams’ workload and speeding up request handling.

See also: Unified Inbox, Human handoff, First Contact Resolution (FCR) rate

XXVIII

Chat widget

A chat widget is a small interactive interface, usually embedded in the corner of a website or app, that lets visitors start a real-time conversation with an agent (human or AI). It often appears as a speech-bubble icon. It makes it easier to reach customer support, product information, or navigation help, offering instant, personalized communication. Modern chat widgets can be enhanced with features such as dynamic FAQs, contact forms, or integration with customer relationship management (CRM) systems.

Why it matters. Adding a chat widget to your e-commerce site greatly improves the customer experience by offering immediate support. It lets you answer questions in real time, guide purchases, and reduce cart abandonment, turning visitors into satisfied buyers thanks to the efficiency of an AI agent.

See also: AI agent, Conversational AI, Omnichannel support