Trend Analysis¶
Visualize time-series data, detect anomalies, and apply seasonal adjustments to understand how value drivers behave over time. Trend Analysis helps you distinguish between signal and noise in your value data.
Who this is for¶
End User Admin Executive
Prerequisites¶
- Multiple periods of actuals or forecast data.
- Reviewed Core Concepts: Forecasts.
- Reviewed Core Concepts: Value Metrics.
Step-by-step instructions¶
1. Open Trend Analysis¶
- Navigate to Analytics → Trends.
- Select a tenant, portfolio, or specific account.
2. Select a driver¶
- Choose a value driver or aggregate metric from the dropdown.
- Available options include all drivers with at least 2 periods of data.
3. Choose a view¶
| View | What it shows | Best used when |
|---|---|---|
| Time-Series | Raw actuals and forecasts on a shared timeline | You need to spot directional shifts |
| Anomaly Detection | Points that deviate beyond expected bounds | You suspect data entry errors or one-time events |
| Seasonal Adjustment | Underlying trend after removing repeating patterns | Your data has quarterly or annual cycles |
4. Set the window¶
- Use the date range controls to focus on specific quarters or years.
- Preset ranges: Last 3 Months, Last 6 Months, Last Year, All Time.
- Custom ranges are supported for up to 24 months.
5. Interpret anomalies¶
- Hover over flagged points to see:
- Expected range (mean ± 2 standard deviations)
- Actual value
- Deviation percentage
- Click a flagged point to open the Realization Plan for that period.
6. Export the chart¶
- Click Download PNG for presentations.
- Click Download CSV for further analysis in Excel or Python.
Displays raw actuals and forecasts on a shared timeline. Useful for spotting directional shifts. Forecasts are shown as dashed lines; actuals as solid lines.
Flags points outside ±2 standard deviations. Useful for catching data entry errors or one-time events. Anomalies are red diamonds; normal points are blue circles.
Applies a decomposition model to isolate trend, seasonality, and residual. Useful for recurring benefits. The adjusted series removes seasonal peaks and troughs.
Tip: Combine with Benchmarking
Overlay peer median data from the Benchmarking module to see if your trend aligns with industry patterns. If your cost reduction trend is flat while peers are declining, you may have a hidden opportunity.
Permissions required¶
| Role | Permission | Scope |
|---|---|---|
| User | View / Export | Assigned initiatives |
| Admin | Configure detection sensitivity / Seasonal models / Set cycle lengths | Tenant-wide |
| Executive | View / Export | Portfolio |
Limits and guardrails¶
Limit Anomaly detection requires at least 6 data points. Limit Seasonal adjustment requires at least 2 full cycles of data. Limit Trend data is retained for 24 months.
Troubleshooting¶
Issue: Anomaly detection flags every point
Cause: The sensitivity threshold is too low, or the data has high natural variance. Resolution: Increase the sensitivity threshold in Analytics → Trends → Settings, or smooth the data with a moving average. Check if a driver was redefined mid-series.
Issue: Seasonal adjustment looks flat
Cause: The data does not contain a strong seasonal pattern, or the cycle length is misconfigured. Resolution: Verify the cycle length (for example, 4 for quarterly, 12 for monthly) and ensure you have at least 2 full cycles. Try the Time-Series view first to visually confirm seasonality.
Issue: Trend data stops at a certain date
Cause: Actuals were not entered for recent periods, or the account was archived. Resolution: Enter missing actuals in the Realization Plan tab, or check the account status in the Accounts list.
Issue: Export CSV has missing columns
Cause: Some periods have no forecast or actual value for the selected driver. Resolution: This is expected. Fill missing cells with your own interpolation if needed.
Related pages¶
- Forecast Analytics
- Benchmarking
- Value Realization Analytics
- Interpreting Analytics
- Core Concepts: Forecasts
Escalation path¶
For calculation errors or missing trend data, open a support ticket with severity P4.