Category: Product Updates

A manual timesheet is a memory test disguised as a report. At the end of the week, someone tries to reconstruct five days of context-switching from memory, rounds every entry to the nearest half hour, and submits a document that looks precise but rarely is.

What a Confidence Score Actually Measures

Instead of asking someone to remember their day, confidence-scored tracking watches the real signals as they happen: keystroke cadence, mouse movement, app and window focus. Every activity block gets a 0–100 score based on how confidently the system can say real work was happening — not how long a timer happened to run.

• No manual start or stop — tracking begins the moment work does
• Every block scored independently, so one quiet hour doesn’t blow up a whole day’s number
• Short gaps between active sessions merge automatically into one continuous block

A timesheet answers ‘what do you remember doing?’ A confidence score answers ‘what actually happened?’ Those are rarely the same number.

Why This Matters for Billing

For any team billing clients by the hour, this distinction has a direct dollar value. A rounded, remembered timesheet is the first thing a client questions during an audit. A confidence-scored activity log, tied to real input signals, is the first thing that ends the conversation.

Every block is scored the same way, whether it’s billed to a client or reported to a manager. Explore Time Tracking →

“Burnout score” sounds like it should be squishy — some vague sentiment pulled from a survey nobody filled out honestly. TrackDots’ version is the opposite: a rolling calculation built entirely from real, observable activity over the last 14 days.

The Six Inputs

• Long hours — days that consistently run past a configured threshold
• Missed breaks — no meaningful pause across the day
• Weekend work — activity on days that should be off
• Late nights — work outside a configured normal-hours window
• Declining productivity — a real drop against that employee’s own baseline
• Weekly hours over limit — sustained overload, not a single busy week

Each signal is evaluated independently and shown that way — an employee’s flag might say “long hours, 6 days” and “late nights, 1 day” rather than a single blended number nobody can unpack.

Four Tiers, Recalculated Daily

Every employee lands in High, Medium, Watch, or All Clear, based on how many of the six signals are currently triggered. Because the window rolls forward every day, a single rough week doesn’t linger as a permanent red flag — the score reflects the last two weeks, not an entire quarter.

A rolling window means the score describes right now — not a bad sprint from six weeks ago that everyone’s already recovered from.

Every High or Medium flag comes with a plain-language recommendation, not just a number. See Burnout Detection →

Configurable, on Purpose

What counts as a “long day” or a “late night” varies by team, season, and culture. Every threshold behind this score is configurable per organization — and changes only ever apply going forward, never rewriting how past weeks were scored.

Most teams don’t lose track of one employee at a time — they lose track of dozens, a little at a time, until a pattern that was obvious in hindsight goes unnoticed for weeks. Anomaly Detection exists to catch those patterns while there’s still time to do something about them.

Four Patterns, Flagged Automatically

• No-activity streaks — ten or more consecutive weekdays with zero tracked activity
• Sudden hour drops — a day’s hours falling sharply below someone’s own personal average
• Erratic start times — a daily start time swinging widely across recent active days
• Unusual working hours — work logged unusually early or late relative to someone’s normal pattern

Every flag is ranked Critical, Warning, or Info, so a manager reviewing a team of twenty doesn’t have to treat every flag with the same urgency.

The goal isn’t to flag everything. It’s to flag the handful of things a manager watching twenty dashboards would otherwise miss.

Evidence, Not Verdicts

Every anomaly links directly to the employee’s real activity diary and profile — nothing is presented as a conclusion on its own. A flag is a prompt to look closer, backed by the actual data behind it, not an automated judgment about anyone’s performance.

Flags are ranked by severity and link straight to the underlying activity data. See Anomaly Detection →