Marketing glossary

Preparing content for answer engines: the answer up front, a question-led structure and machine readable data. The goal is to be the source of the answer, not an entry in a list.

A purchase carried out by an AI agent on the user’s behalf: it finds the product, compares offers and often completes the order. What decides is not the look of the shop but the quality of the product data.

The crawlers of language models that collect content for training and answers. When a host or firewall blocks them, the brand stops existing for that AI, and usually nobody knows.

Google’s conversational search mode, where the user asks follow-up questions and receives a continuous answer instead of a list of links. Sources are cited inside the answer, so a brand can be highly visible and still receive no clicks at all.

AI generated answers shown above Google’s organic results. The user gets a summary with source links, so a brand can be highly visible without earning a single click.

A brand’s measurable share in AI assistant answers: how often it is mentioned, whether it is cited and in what wording. Not a single number, but a set of indirect indicators.

The average value of an order. Raising it is usually a cheaper route to growth than buying more traffic, because it monetises visits that are already paid for.

ROAS calculated from total revenue and total ad spend across channels. It shows the company’s reality, not each platform’s optimistic numbers.

The ROAS at which a campaign neither earns nor loses. It is derived from margin, not revenue, and without it no target ROAS has any basis.

The cost of acquiring one new customer: total marketing and sales cost divided by the number of new customers. It only means something paired with LTV.

Product results inside ChatGPT. Visibility follows structured data, feed quality and what is published about the product, not the size of an auction bid.

The share of customers lost in a period. In subscription models it decides whether acquisition makes sense at all, because it sets how long a customer keeps paying.

A brand’s share of citations in AI answers on a topic. The share of voice metric for an era where a model answers instead of a list of links.

Revenue after variable costs. The amount that pays for advertising and fixed costs, and the basis for calculating POAS and break-even ROAS.

Cost per conversion: what one order, lead or sign-up costs. Unlike CPC it measures the outcome, not the visit, but without margin it is still half the story.

Keeping customers and making them buy again. A cheaper growth lever than acquisition, because it works on customers who are already paid for.

Experience, expertise, authoritativeness and trust. The frame Google and AI models use to judge whether a source is worth citing at all.

Optimising for entities, meaning uniquely identified things and brands, rather than keyword strings. Search engines and language models reason in entities, not words.

Optimizing content so generative search engines and AI assistants cite it in their answers. The goal is not a position in a list of links but a mention at the moment someone is deciding.

An incrementality test that switches advertising on or off in selected regions. The gap against control regions shows the campaign’s true contribution.

A control group deliberately excluded from advertising. The behavioural gap between the groups shows how many conversions would have happened without the campaign.

The revenue lift that would not have happened without the campaign. It separates advertising that caused the result from advertising that merely claimed a conversion that was coming anyway.

A protocol for actively notifying search engines about content changes instead of waiting for a crawl. For Bing, and therefore for ChatGPT and Copilot, it is the fastest route into the index.

A database of entities and their relations from which Google draws facts about companies, people and concepts. When a brand is missing, AI assembles its facts from elsewhere.

A text file at the site root that tells language models who you are and which content matters. It is not a standard enforced by search engines, it is a curated index you write yourself.

The value of a customer over the whole relationship, ideally in margin, not revenue. It sets how much you can afford to pay to acquire one.

Customer value divided by acquisition cost. Below 3 the model usually gets fragile, well above 5 often means growth is being throttled unnecessarily.

A statistical model that estimates each channel’s contribution to revenue from historical data. It needs no cookies and no user tracking, which is why it is back after measurement tightened.

Total revenue divided by total marketing spend. An account-wide view that ignores attribution disputes, which is why attribution cannot fool it.

The time it takes to recover the cost of acquiring a customer. It decides how fast a company can scale without outside funding.

A search-first AI assistant that answers with visible sources. Relevant for brands because its citations are transparent and trackable.

Profit per unit of ad spend. Unlike ROAS it accounts for margin, so it separates revenue that earns from revenue that merely flows through.

The method where a language model retrieves sources before answering and builds the answer from them, instead of drawing only on what it learned in training. RAG is the reason AI visibility can be influenced by content.

Search based on the meaning of a query rather than word matching. It is why one page gets found through dozens of differently worded queries, and why writing for one exact keyword stopped working.

Marking up content with the schema.org vocabulary so machines know what a page holds: a service, a product, a question, a company. The basis for being understood correctly by search engines and AI.

A search that ends without a click, because the user gets the answer directly in the results. AI answers push its share up and change what should be measured as success.