Estimate Accuracy: The Research

Metacognition, the ability to monitor and evaluate your own cognitive processes, was formalized by Flavell (1979) as a distinct domain of cognitive development. It includes self-monitoring ("how am I doing?"), self-evaluation ("was my estimate accurate?"), and self-regulation ("what should I adjust next time?").

Barkley, Murphy, and Fischer (2008) documented that ADHD adults significantly overestimate their own performance and underestimate task duration. This isn't optimism. It's a measurable metacognitive deficit: the internal feedback system that tells you "this is taking longer than you thought" doesn't fire accurately.

Toplak, West, and Stanovich (2013) showed that self-report measures of executive function and performance-based measures diverge significantly in ADHD, confirming poor metacognitive calibration. ADHD individuals think they're performing better than they are, and think tasks took less time than they did.

Ebbi's estimate accuracy feature tracks planned versus actual duration for every task. Over time, it shows patterns: "your tasks usually take 35% longer than planned." This is external metacognitive feedback that the ADHD brain struggles to generate internally.

The goal isn't to make users feel bad about estimating poorly. It's to make the calibration error visible so it can be corrected. A user who consistently underestimates by 30% can start adding a buffer. A user who overestimates can recognize that tasks aren't as daunting as they seem. Both are improvements in self-awareness that directly improve planning accuracy over time.

This feature is gated behind Pro because it requires session history data to compute meaningful patterns. Free tier users build the data during their trial and see the benefit, creating a natural upgrade incentive tied to genuine value.

No study has directly tested whether tracking planned-vs-actual time in an app improves future planning accuracy for ADHD adults. The metacognitive deficit is well-documented. The intervention (external feedback to correct calibration) is theoretically grounded. The specific implementation has not been validated in a controlled trial.

Ebbi's usage data may eventually provide evidence for this approach. Until then, we are transparent about the gap between the supporting research and the specific feature design.