Confluence

Public discussion of Confluence across the four-week window was heavily dominated by frustration, with complaint-coded mentions outnumbering praise-coded ones by a wide margin. Bug reports were the single largest theme by a considerable distance, followed closely by reliability concerns and missing features, suggesting that day-to-day functional grievances were the primary force shaping how commenters talked about the product. Praise did surface, mainly around specific features and integrations, but the volume was modest enough that it did little to counterbalance the negative current running through the broader conversation.

Sentiment showed a modest upward movement in the middle weeks, climbing from the opening baseline and reaching a relative high around mid-May, when mention volume was also at its lowest. That pattern suggests the more positive stretch may have reflected a quieter, self-selecting slice of discussion rather than a genuine shift in satisfaction. As mention volume surged sharply in early to mid-June, sentiment moved back down and settled near its lowest point of the tracked window, implying that whatever drew more commenters into the conversation late in the period skewed negative.

A recurring and pointed thread in the sample mentions concerned Atlassian's announced use of customer data for AI training, with commenters expressing unease about opt-out limitations and the scope of affected organizations. This data-use question appeared to carry real emotional weight, with some discussion explicitly framing open-source alternatives as a counter-position. The forced AI-generated page titles and emoji-heavy AI writing assistance also drew irritated commentary, reflecting divided opinion on AI integration specifically: a small number of mentions treated it neutrally or curiously, while others found it presumptuous or disruptive to existing workflows.

Opinion was most visibly split around AI features overall. Some commenters engaged with tools like Rovo with dry humor rather than outright rejection, while others voiced direct resistance to AI acting on their content without explicit prompting. The reliability and bug themes attracted little visible disagreement, trending toward shared exasperation rather than debate.