What is a confusion score?
A confusion score is a per-page metric that rolls the frustration signals on that page, weighted by severity and reach, into one number tracked against the page's own baseline. It exists so friction can be ranked, alerted on, and compared before and after deploys.
A confusion score is one number per page that answers "how much are users struggling here right now." It rolls up the page's frustration signals, the rage clicks, dead clicks, thrash, loops, and form struggles, weights them by severity and by how many sessions they touched, and compares the result to the page's own history.
The point of the score is not the number itself. It is what a single comparable number makes possible.
Why one number
Raw signals answer "what happened on this element." They do not answer the questions a team actually schedules work around:
- Which page deserves attention first? Ranking requires a common scale. A score lets a busy pricing page and a quiet settings page be compared fairly, because each is measured against its own normal.
- Is this getting worse? A score is a time series. Trend, not snapshot.
- Did the deploy break something? Score before, score after. When confusion on a page jumps within hours of a release, the release is the suspect, with a timestamp.
- Did our fix work? The same comparison in reverse, the verification step most teams skip. A fix that worked shows up as a score that fell and stayed down.
Against its own baseline
The design decision that makes a score honest is the baseline. Pages differ: a complex dashboard produces more raw signal than a landing page without being more broken. Scoring against the page's own history separates "this page is busy" from "this page is worse than it was." Alerting keys off the deviation, so quiet pages that suddenly spike get attention even when their absolute numbers stay modest.
What a score cannot do
A score locates and ranks; it does not diagnose. The number says page X got worse Tuesday. The diagnosis lives in the signals underneath: which element, which behavior, which sessions. A good friction system uses the score for triage and keeps the evidence attached, so the drill-down from "this page is red" to "this button, these sessions" is one step.
Flusterduck computes confusion scores continuously for every tracked page and ties them to deploys, so verification happens automatically. A free scan produces a first score for your site's key pages today.