onemore-telemetry-extrapolation

# OneMore Telemetry Extrapolation Estimate how often each OneMore command is executed across your entire user base, even though only a subset of users have opted into telemetry. ## Overview Your telemetry data captures only a fraction of actual usage (opt-in users). This skill extrapolates from known telemetry counts to estimate total command executions across all users by modeling different opt-in rate scenarios. **Key assumption:** Telemetry users behave similarly to non-telemetry users (no selection bias). ## Methodology ### Input Data Required 1. **Unique telemetry sessions per month** — e.g., 3,368 2. **Monthly downloads** — e.g., 2,992 3. **Top command counts from telemetry** — with per-month averages for each command The maintained Excel file containing this data is here: "C:\Users\steve\OneDrive\OneMore\Metrics.xlsx" ### Calculation For each command, divide the telemetry count by the assumed opt-in rate: ``` Estimated Total Usage = Telemetry Count ÷ Opt-in Rate ``` **Example:** If ExportCLI shows 7,249 executions in telemetry, and you assume 10% opt-in: - Estimated total = 7,249 ÷ 0.10 = **72,490 executions/month** ### Scenario Models | Scenario | Opt-in Rate | Use Case | |----------|------------|----------| | **Conservative** | 5% | Very low adoption; privacy-conscious user base | | **Moderate** | 10% | Balanced assumption; typical default behavior | | **Optimistic** | 20% | Higher adoption; users generally accept telemetry | | **Very High** | 30% | Aggressive opt-in; enthusiast or enterprise users | **Most likely:** 10-20% opt-in, based on typical SaaS adoption patterns. ## Usage ### Step 1: Gather Your Data Collect from your telemetry system: - Total unique sessions recorded this month - Average sessions per month (over your tracking period) - Top N commands, with counts and per-month averages - Download volume (context only) ### Step 2: Choose Your Scenario Pick the opt-in rate that best matches your user base: - **Opt-in telemetry is required or pre-enabled?** → Use Very High (30%) or Optimistic (20%) - **Opt-in is default but users can disable?** → Use Moderate (10%) - **Opt-in requires deliberate action?** → Use Conservative (5%) ### Step 3: Scale the Data For each command, multiply by the inverse of your chosen opt-in rate: ```python estimated_usage = telemetry_count / opt_in_rate ``` Python example: ```python commands = { 'ExportCLI': 7249, 'ApplyStyle': 4780, 'PasteText': 1811, } opt_in_rate = 0.10 # 10% scenario for command, count in commands.items(): estimated = count / opt_in_rate print(f"{command}: {estimated:,.0f}/month") ``` ### Step 4: Analyze Results - **Top commands** tend to be used 10-100x more often than edge commands - **Power-user commands** (e.g., CLI, batch operations) dominate usage - **Formatting/styling commands** are second-tier staples - **Niche/specialized commands** cluster at the low end ## Example Results Based on OneMore data (May–Sep 2026): | Command | Telemetry | Moderate (10%) | Optimistic (20%) | |---------|-----------|---|---| | ExportCLI | 7,249 | 72,490 | 36,245 | | ApplyStyle | 4,780 | 47,803 | 23,902 | | PasteText | 1,811 | 18,110 | 9,055 | | InsertToc | 1,709 | 17,087 | 8,543 | | JoinParagraph | 1,534 | 15,340 | 7,670 | **Interpretation:** In the Moderate scenario, ExportCLI is run ~72k times per month across all users—roughly 2.5x more than ApplyStyle. ## Important Limitations ⚠️ **Selection bias.** Users who opt into telemetry may not represent the full user base: - Power users might opt in more (inflating active-command estimates) - Privacy-conscious users opt out (missing conservative usage patterns) - Different segments (free vs. paid, enterprise vs. individual) may have different opt-in rates ⚠️ **Downloads ≠ active users:** - Same user downloading updates multiple times - Downloads from users who never activate the product - No data on churn (how long users remain active) ⚠️ **Sessions ≠ users:** - A single user may have multiple sessions (different machines, reinstalls) - No visibility into one-time users vs. power users ⚠️ **Data quality:** - Telemetry may miss certain commands (network issues, offline mode) - Commands may be recorded multiple times per action - Session lifetime definition affects session counts ## When to Use This Analysis ✅ **Good for:** - Prioritizing which features to optimize (focus on high-usage commands) - Forecasting server load or backend traffic - Identifying feature tiers (which commands power users rely on) - Comparing relative usage (ExportCLI vs. ApplyStyle) more reliably than absolute numbers - Detecting seasonal or trend changes in command popularity ❌ **Not reliable for:** - Precise user counts (opt-in rates vary too much) - Predicting user growth (downloads are lagging, not predictive) - Per-user metrics (no session-to-user mapping) - Business metrics (revenue, churn, retention) ## Refining Your Estimate Over time, you can calibrate your opt-in rate: 1. **Run an A/B test:** Disable telemetry, track product usage another way (server logs), then compare 2. **Survey users:** Ask opt-in vs. opt-out users about their behavior 3. **Compare with server logs:** If you have backend data (API calls, file writes), correlate with telemetry counts 4. **Monitor anomalies:** If a command's telemetry count suddenly spikes/drops, check if opt-in rate changed ## Tools & Scripts ### Python Script ```python import json def extrapolate_usage(telemetry_data, opt_in_rate): """ Scale telemetry counts to estimate full-user-base usage. Args: telemetry_data: dict of {command: telemetry_count} opt_in_rate: float (e.g., 0.10 for 10%) Returns: dict of {command: estimated_total} """ return { cmd: count / opt_in_rate for cmd, count in telemetry_data.items() } # Usage telemetry = { 'ExportCLI': 7249, 'ApplyStyle': 4780, 'PasteText': 1811, } scenarios = { 'Conservative (5%)': 0.05, 'Moderate (10%)': 0.10, 'Optimistic (20%)': 0.20, 'Very High (30%)': 0.30, } for scenario_name, rate in scenarios.items(): estimated = extrapolate_usage(telemetry, rate) print(f"\n{scenario_name}:") for cmd, est_count in sorted(estimated.items(), key=lambda x: -x[1]): print(f" {cmd}: {est_count:,.0f}/month") ``` ### CSV Format Export your results to CSV for spreadsheet analysis: ``` Command,Telemetry,Conservative (5%),Moderate (10%),Optimistic (20%),Very High (30%) ExportCLI,7249,144980,72490,36245,24163 ApplyStyle,4780,95607,47803,23902,15934 PasteText,1811,36220,18110,9055,6037 ... ``` ## References - **Opt-in telemetry best practices:** https://telemetry.microsoft.com/ - **Selection bias in observational data:** https://en.wikipedia.org/wiki/Selection_bias - **Survivorship bias & telemetry:** https://www.lessig.org/survivorshipbias/ --- **Last updated:** September 2026 **Methodology source:** Extrapolation from OneMore telemetry, May–September 2026