Measuring

Presence, activity and output volume are captured continuously and predict little. Manager quality, whether people can speak up, and whether trajectory is visible predict a great deal and appear on no dashboard.

A Stanford study in Nature found no productivity drop under hybrid working and 33% lower turnover (peer-reviewed); Gallup finds remote workers logging about an hour less per day with steady output. Hours were never what was being purchased.

An activity score is built from input-device events, application categories set by default, and an idle threshold. It misses thinking, conversation and paper work by construction, and the work that scores highest is the most automatable. For a related reference, see read more.

Six things predict retention and none arrives automatically: turnover by manager, whether people can speak up, whether trajectory was discussed, feedback recency, internal movement, and departure clustering. Together they cost roughly two days a year for an organisation of a hundred.

Your own turnover data, cut four ways, answers more than any purchased research: by manager, by tenure, by date against events, and by whether you minded.

And seven things cannot be instrumented at all — a known blind spot being safer than a false reading.

For additional context on this topic, see Pixlr.