Prompt
A prompt is the question or instruction a person types or speaks to an AI system, such as a chatbot or an AI search assistant, which the model reads to decide what to retrieve, generate, and cite in its answer.
Prompts are the new keywords. Where shoppers once typed a few terse search words, they now ask AI assistants full questions in plain language: which option suits a small kitchen, which model lasts longest, which brand people actually trust. The phrasing is longer, more specific, and closer to how someone would speak to a knowledgeable friend, so the surface a buyer reveals about their intent is far wider than a keyword box ever captured. A prompt usually carries three things at once: the subject the shopper cares about, the constraint that narrows the field (budget, size, skin type, climate), and an implied standard for what a good answer looks like. Reading all three, rather than just the nouns, is what separates content the model can use from content it scrolls past.
The practical move is to phrase content as direct answers to the prompts your buyers actually use. Lead with the answer in the first sentence, name the product or use case plainly, and let the supporting detail follow. Models tend to lift a clean, self-contained statement that resolves the question, so a paragraph that buries the conclusion three sentences down is less likely to be quoted than one that states it up front.
Consider a Shopify store selling cast-iron cookware. The owner notices that buyers in chat assistants rarely ask for a product name; they ask things like which pan works on an induction hob, or whether a skillet is too heavy for someone with a weak wrist. So she rewrites her collection copy and product FAQs to answer those exact questions in plain sentences: the 24cm skillet is induction-compatible and weighs 1.4kg, lighter than the 28cm pan many first-time buyers reach for. She also organises customer reviews so the ones mentioning weight and hob type sit near the top. The page now reads as a set of small, quotable answers rather than a brochure.
This matters for AI search because assistants like ChatGPT, Perplexity, and Google AI Overviews resolve a prompt by stitching together passages that cleanly match the question, then citing the sources behind them. A page written to answer real prompts gives those systems something precise to quote and attribute, which is how a store earns a mention inside an answer instead of being summarised away. The honest caveat is that you cannot see every prompt a customer types, and AI systems paraphrase questions internally, so treat prompt-matching as informed inference rather than a precise targeting tool. Map the genuine decisions behind a purchase (fit, durability, comparison, trust) and answer those, instead of chasing exact wordings you can only guess at.