How we calculate your energy trend.

Recent work counts more.

We use a 3-day half-life. An hour rated today has twice the influence of an hour rated three calendar days ago, and four times the influence of one rated six days ago.

Weight = 2 ^ (−days ago / 3)

Score = 100 × weighted gave-energy hours
              ÷ all weighted rated hours

We apply those weights to each day’s rated hours. Unmarked time stays out of the calculation. Twelve five-minute entries count the same as one hour with the same rating.

A score of 0 means all the weighted rated hours took energy; 50 is an even mix; 100 means all gave energy. These describe the mix of rated work.

The trend line and positive-energy percentages share a continuous colour scale: red and orange at the low end, blue around 50%, and increasingly green toward the high end. Matt Mochary’s energy-audit guidance recommends spending 75–80% of your time on energising work. The colour bands are our product guidance, not measures of performance or health thresholds.

Better days shift the trend.

Ten mixed days mostly take energy, then the balance shifts: more work gives energy, with a few fully green days. Each day has eight rated hours. Two earlier rated days establish the starting trend.

The trend responds to the daily ups and downs without forgetting the preceding stretch. This is a calculated example, not a prediction of how a person will feel.

The line and bars tell different parts.

The line is the recent weighted score. The bars and breakdown table show actual logged hours: gave, took and unmarked. A completely green day can therefore have a recent trend score below 50 because earlier red days still contribute.

On a rated day, only that day and earlier work contribute to its score. Changing the chart’s date range does not reset the calculation. Rating old time updates the day the work happened and the trend after it.

Connections across missing ratings.

The line starts with your first rating. Solid sections show scores on rated days. A lighter dashed line connects the scores across days with no ratings, using a straight interpolation. These in-between values are inferred, not measured, and never count toward rated hours or the underlying score calculation.

After the latest rating, a dashed line holds the last known score and shows when it was rated. When new feedback arrives, that connection updates to meet the new score. We do not assume a gap means recovery or decline.

Every rating contributes immediately. Early scores can change substantially as you rate more time. Selectively rating only some work can still skew the result.

Every person gets equal weight.

For a team or company, we calculate each eligible person’s score first, then average them equally. Someone logging longer hours does not dominate the line. A person contributes from their first rating. When calculating a new team score, people without a rating in the last 7 days are excluded, not counted as zero. Between rated days, the chart connects team scores rather than changing the line just because someone’s rating became older.

The people contributing can change over time, which can also move the line. Client, project, task and label breakdowns remain based on hours, so their overall split may differ from the team score. Filters also apply to the historical work used for the trend. An “all of” filter across multiple clients, projects, tasks or labels includes days where that person worked on every selected value.

Members see their own ratings. Managers can see members of their assigned teams; company admins can see the organisation. Only you can rate your time.

Research-informed. Still a product model.

Exponential smoothing is an established statistical way to give recent observations more weight. Rutledge and colleagues’ 2014 research used discounted past events to model momentary happiness in a decision task. It supports exploring recency weighting, but does not validate this work-energy score or a three-day decay.

Subjective vitality research measures how energetic people feel directly. Work and recovery research also considers what happens outside work. Our task ratings don’t capture all of that, which is why we don’t label the score a battery estimate.

The three-day half-life is our initial design choice. We have not validated it as a predictor of people’s perceived energy. Future changes should be tested against direct, voluntary self-reports, rather than chosen solely because the chart looks convincing.

Model: work-energy-v2 · Three-day half-life · Equal-person team average