A small dataset.
A complete chain of evidence.
Serein includes 56 fixed sample days, from July 7 through August 31, 2026. Each day has a new-visitor record and a returning-visitor record. Values are produced by a deterministic formula with a trend and a weekly rhythm. They are not measurements from a real product.
The fixed period makes the demo repeatable. Selecting the final seven, fourteen, or twenty-eight sample days always uses the same endpoint. The matching comparison is the preceding period of equal length.
Learning how a segment, period, and metric change an analysis. Practicing a clear observation and keeping the next question attached.
The denominator matters.
| Metric | Calculation | What it means |
|---|---|---|
| Average daily visitors | Sum of visitor-days ÷ selected days | A typical day's volume, not unique people across the period |
| Total product events | Sum of events in selected records | Total illustrative interactions during the selected period |
| Activation rate | Total completions ÷ total starts × 100 | A weighted ratio of the two sample activation steps |
Activation is calculated from the combined numerator and denominator. Averaging daily percentages would give small and large days equal influence and could produce a different result.
All visitors combines the two segment records for each day before calculating the chosen metric. The chart, summary, comparison, funnel, and table use the same selected rows.
A change is an observation.
Not an explanation.
Visitor and event comparisons show relative percentage change from the preceding equally sized period. Activation comparisons use percentage-point change. A move from 50% to 55% is five percentage points; it is not described as a five-percent relative increase.
The chart begins at zero. Activation uses a fixed zero-to-100 scale. Count charts use a rounded upper bound above their highest point. These choices keep the visual interpretation explicit.
The dataset has no uncertainty model, experiment assignment, or real behavioral context. Serein therefore makes no causal or statistical-significance claim. A difference between segments suggests a question to investigate, not a reason to declare a winner.
Inspect the calculationSave the lens.
Write the interpretation.
A saved view stores its name, date-range length, segment, metric, and chart style in this browser. The fixed dataset recreates the same view each time. Your note remains separate from the calculated values.
The notebook holds up to twelve views. Notes save locally as you type. If storage is unavailable, the interface reports the failure and keeps your current note on the page. Exporting a report includes the controls, summary, note, and underlying daily data as Markdown and CSV text.
No tracking script, external analytics account, uploaded dataset, or server-side profile is involved. The demo is a self-contained place to explore the relationship between data and a useful question.
Open the notebook