Bot Detection Criteria: Days where (new_users / sessions > 0.85) AND (bounce_rate < 0.20) are flagged as bot-inflated. This identifies automated traffic that navigates multiple pages (low bounce) but generates no returning visitors (high new-user ratio). The same criteria are used in the live analytics dashboard.
Session Data Source: maggies-analytics.staging.ga4_daily_metrics — daily GA4 session data exported natively to BigQuery.
Revenue Data Source: marts.daily_summary (BigCommerce DTC orders). Nov/Dec 2025 capped at ~$200K to exclude wholesale/Black Friday bulk. Raw Nov = $1.2M, raw Dec = $311K.
2026 Forecast Source: forecast_2026.csv — the targets currently used for contract evaluation (sessions and revenue).
2026 Actuals: Full-month data for January and February. March is partial (revenue: 5 days, sessions: 4 days as of March 5).
Clean Sessions: Total sessions minus bot sessions for each month. Only February 2025 had detectable bot activity; all other months' clean and raw totals are identical.
Limitations:
- Bot detection is heuristic-based. Some bot traffic may not match these criteria, and some edge-case legitimate traffic spikes could theoretically be misclassified (though the Feb 2025 pattern is unambiguous).
- Only February 2025 triggers the bot detection filter. If other months had subtler bot inflation, their "clean" numbers may still be slightly overstated.
- Growth rate projections assume steady YoY improvement; actual performance will vary by seasonality and campaign activity.
- Revenue figures exclude wholesale and marketplace channels. Nov/Dec 2025 are DTC estimates (~$200K each) due to Black Friday and wholesale volume in the raw data.