Loading
Because the writing queue is guesswork. Somebody picks topics that seem important, and the articles that result answer questions nobody was asking in words nobody was using.
The failed search report lists what people typed into help centre or widget search and found nothing useful.
Every search with no result, or with results nobody opened, is recorded and grouped by phrase. The report is ordered by frequency, so the queue writes itself.
A phrase is searched sixty times in a month with nothing returned. That is the next article, and it is written in the customer's words rather than the product's.
It will not tell you the answer. It tells you the question is being asked and going unanswered; somebody who knows the product still has to write it.
It feeds the knowledge base backlog, and can be compared against ticket causes to show where a missing article is generating contact.
Yes, so forty wordings of one question read as one item.
Yes, separately — agents failing to find something is its own signal.
Yes, so the cost of the gap is visible.
Fourteen days, every module, no card. Or half an hour with someone who will run it on your own records and tell you where it does not help.