Context and document match – what does it mean?

How it works

When a visitor asks a question, the assistant first searches the company’s own knowledge base. From the texts of the website, the product pages, the price list and the documents, it filters out the few paragraphs that best fit the question.

It then formulates the answer from this excerpt, and the source is shown next to the answer. If there is no good enough match, the assistant does not guess: it asks for the contact details and hands the question over to a colleague.

An example at a company

The customer asks how long delivery to Austria takes. In the knowledge base, the assistant finds the paragraph on international shipping and formulates the answer from it.

The source also appears next to the answer, so the customer can check it. If the company has not written about this, the question goes to a colleague.

What this means for business leaders

The quality of the answer depends on the quality of the match. If the topic is not in the knowledge base, there will be no good answer either, even though the question is real. So check that the product pages, the price list and the frequently asked questions have been uploaded.

Ask the provider whether the source is shown next to every answer, and what happens when there is no match. Because of an incomplete knowledge base, the customer may get a partial or inaccurate answer.

How we do it

The assistant answers from your website, product pages, price list and documents, and the source is shown next to the answer. What there is no content for, it does not make up: it asks for the contact details and hands the question over to a colleague. Onboarding is done by our team, and before go-live we check it with ten test questions.