Long-tail keywords are longer, specific queries with low search volume but clear intent – e.g. "ux audit online store cost". Individually they bring little traffic; added together they often bring more than generic phrases, and they convert far better.
A generic phrase ("websites") has high volume, fierce competition and unclear intent: is the person looking to order a site, learn to build one, or find a definition? A long-tail phrase ("how much does a company website cost") says exactly what the user needs – so it's easier to write content that hits, and easier to convert.
The long tail has found a second life in AI search: people ask assistants full questions, not keywords. A site built around real customer questions (question headings, answer-first paragraphs) serves both worlds at once.
In practice: on a limited budget, the long tail is the shortest route to customers. Instead of fighting for one giant phrase, you build a set of pages and articles answering specific buying questions. That is how the "one service = one page" structure works.