Insight tags are nouz's way of saying "this is worth your attention". Each one fires from a specific statistical rule — knowing the rules helps you read the signals correctly.
01 The full tag list
- Peak day — your highest-revenue weekday, consistently above average.
- Weak day — your lowest-revenue weekday, consistently below average.
- Streak — a run of consecutive days above (or below) a threshold.
- Margin drift — a product's per-unit margin shifting more than ~10% over four weeks.
- Best week — the current week is your top-revenue week of the period.
- Cost creep — a category of expense growing faster than revenue.
- Cap warning — approaching 80% of your plan's annual revenue cap.
02 What triggers each
Each tag has a threshold + a confidence window. Tag rules live in lib/statistics-rules.ts if you're curious about exact values. The general approach: median + coefficient-of-variation filtering, so one weird Tuesday doesn't flip a long-term pattern.
03 Why we don't fabricate
Most analytics dashboards manufacture insights to fill the page — "your revenue is up 3.4% this week!" — even when 3.4% is well within noise. We refuse. If the signal isn't strong enough to be honest, the Insights panel stays empty. See When insights are quiet for the philosophy.
04 Which to act on first
Some tags are informational (peak day — useful context, no action needed today). Others are actionable (margin drift — fix the underlying issue this week). Rough priority:
- Margin drift / cost creep — act this week. Both signal silent margin loss.
- Cap warning — act this week. Approaching your plan cap.
- Streak (negative) — look closer. Why three bad days in a row?
- Peak day / weak day / streak (positive) — informational. Use for staffing.
Was this article helpful?
Your vote helps us decide what to write next.