Building a grow log and KPI dashboard

A grow log is the continuous, time-stamped recording of input variables (climate, irrigation, feeding) and outcome variables (growth, losses, harvest yield) over a growing cycle.

A KPI dashboard is the condensed view of this log, reduced to decision relevance — not every logged variable deserves a place on the dashboard.

Without timestamps, a symptom that appears later cannot be reliably traced back to a preceding event (e.g. an EC spike or a climate outlier) — the time link is the basic prerequisite for root-cause analysis.

A striking correlation within a single run is statistically weak, because several variables fluctuate at the same time. Only comparison across multiple runs with deliberately held-constant variables allows solid conclusions.

Climate: air temperature, relative humidity, VPD — as daily mean and daily range, not just a snapshot.

Irrigation/nutrients: EC and pH of inflow and drainage, watering amount and interval.

Growth: height, internode spacing, canopy uniformity — as weekly measurement points.

Outcome: dry weight per area, loss rate, and where lab access is available, cannabinoid/terpene content.

A dashboard only becomes truly useful once every displayed metric immediately shows you when you need to intervene.

Trend views (progression over time) are more informative than pure tables of instantaneous values, because they make drift visible before a threshold is exceeded.

For a problem that appears in retrospect, first check the log history for the 5–7 days before the first visible symptom — the triggering deviation almost always occurs before the visible effect, not at the same time as it.

Multiple variables fluctuating simultaneously (e.g. climate failure AND a nutrient change in the same week) make clear attribution difficult — clean logging separates events in time wherever possible.

Tracking too many KPIs without prioritization — the dashboard becomes cluttered, and no single metric gets the attention it needs.

Changing the measurement methodology between runs (different measurement times, different sensor placement), which makes runs no longer comparable.

Treating a single striking correlation from one run as a proven relationship without verifying it in another run.

Statistical control charts from process control can be applied to grow KPIs to systematically separate normal variation from genuine deviations.

Automated sensor logging reduces human recording errors compared to manual logging, but it does not eliminate the need to keep documenting events (interventions, observations) manually and with timestamps.