Knowledge Gap

A knowledge gap is the space between the questions people bring you and the answers your knowledge base can actually give. It surfaces three ways: a search that returns nothing, a ticket topic with no article mapped to it, and — increasingly — an AI assistant that invents an answer because retrieval came back empty. Every unfilled gap is a future ticket; every wrongly-filled one is a future complaint.

There is no canonical number, but teams operationalise the gap as a coverage or failed-search rate: the share of queries that come back with no adequate answering article. You find the specific gaps by mining zero-result searches, tagging tickets that had no reusable article, and logging where an AI assistant fell back to "I don't know" — then turning the top offenders into a monthly "missing articles" list.

The gap is usually a content problem, not a tooling one. Gartner's 2024 self-service survey of more than 5,700 customers found only 14% of service issues get fully resolved online, and 43% of customers who abandoned self-service said they simply couldn't find content relevant to their problem — a knowledge gap by another name.

What it hides: the gap you can measure is only the visible half. Zero-result searches miss the customer who phrased the question in words your content doesn't use and assumed you had nothing, and they can't see the worst gap of all — the article that exists, gets found, and is wrong or out of date, a filled slot that's really empty. In a retrieval-based AI assistant that silent gap is the dangerous one: instead of returning "no results," it returns a fluent, confident hallucination.