How to Use Oracle EPM Intelligent Performance Management (IPM) Features: Complete Hands-on Guide for AI Insights, Forecasting & Performance Analysis

In this guide, I will walk you through how to explore and use IPM (Intelligent Performance Management) features inside Oracle EPM.

Oracle IPM leverages AI and ML to help decision-makers gain predictive insights, detect anomalies, forecast business performance, and automate workflows.

This step-by-step tutorial is designed for EPM users, analysts, and finance teams looking to integrate AI-powered intelligence into their planning and analytics process.

By the end, you will understand how to navigate IPM modules inside EPM, run insights, and interpret results.

Tools Needed

Tool / Item Purpose

Oracle EPM Cloud Access - Access to IPM capabilities

Planning module enabled - For predictive planning steps

Internet browser (Chrome / Edge) - Access EPM dashboard

Your dataset / model - For generating forecasts & insights

Screenshots & video reference - To follow and replicate steps

Access the Oracle EPM Cloud instance and navigate to the dashboard

Sign in to your Oracle EPM Cloud environment (via direct URL or SSO), land on the Home page, and open the Dashboards area where you’ll later launch IPM workflows.

Prerequisites

1. You have a valid EPM Cloud account (or SSO access).
2. Your role includes at least User access to the target application and Dashboard visibility (e.g., Planning: Viewer/Planner/Power User; or equivalent custom roles).
3. Your administrator has granted you access to the dashboards you’ll use.

Detailed steps

1. Open the EPM URL
2. Use your organization’s EPM URL. It typically looks like:
3. https://<identity-domain>.epm.<region>.oraclecloud.com/
4. If you use a bookmarked pod URL (e.g., .../epmcloud or app-specific URL), that’s fine too.
5. Sign in
6. Standard login: Enter your username and password, then click Sign In.
7. SSO (Single Sign-On): You may be redirected to your company’s IdP (Okta/Azure AD/etc.). Complete the SSO prompt.
8. MFA: If enabled, approve the authentication prompt or enter the code.
9. Land on the Home page (Redwood UX)
10. After login, you should see the Home page with cards (tiles) grouped into clusters (e.g., Planning, Dashboards, Data Exchange, Reports).
11. If your organization still uses a classic layout, look for the Navigator (≡) icon—functionality is the same, labels may vary slightly.
12. Open the Dashboards area
13. From Home, do any of the following:
14. Click the Dashboards card directly, or
15. Click the Navigator (≡) in the top-left, then select Dashboards, or
16. Use the Search bar at the top: type “Dashboards” or the specific dashboard name and select it.
17. Choose the target dashboard
18. In the Dashboards list, click the dashboard you need (e.g., “Executive Overview,” “Forecast KPIs,” “IPM Insights”).
19. If your dashboards are organized under an application (e.g., Planning), you may first open that application’s cluster and then pick Dashboards within it.
20. (Optional) Pin or favorite frequently used dashboards
21. Click the star (★) or More menu on a dashboard and select Add to Favorites.
22. On the Home page, click Edit (pencil icon) to pin frequently used dashboards as cards for one-click access.
23. Verify access
24. Confirm the dashboard loads without access errors. If you see “You do not have access” or empty visuals, contact your admin to:
25. Grant the correct role or group membership.
26. Assign dashboard access to your user/group.
27. Ensure required data forms, data grants, or data maps are provisioned.

Open the IPM workspace from the navigation menu.

From the Oracle EPM Home page (Redwood UX), use the Navigator (≡) or Home cards to locate and open the Intelligent Performance Management (IPM) workspace—often labeled IPM, IPM Insights, or Intelligent Insights, depending on how your tenant is configured.

Prerequisites

1. You can log in to Oracle EPM Cloud and reach the Home page.
2. Your role/group is provisioned for IPM (e.g., Planning – Viewer/Planner/Power User or equivalent custom role with IPM access).
3. The IPM feature/tile is enabled and visible in your navigation flow (admin-managed).

Detailed steps

1. Go to the Home page
2. After you sign in, you’ll land on Home with cards (tiles) grouped by function (e.g., Planning, Dashboards, Data Exchange, Reports).
3. Open the Navigator
4. Click the Navigator (≡) icon in the top-left header.
5. The left panel opens, showing a structured list of features and application areas.
6. Locate the IPM workspace
7. Scroll the Navigator and look for one of these common labels (your tenant may use any of these):
8. IPM
9. IPM Insights
10. Intelligent Insights
11. Intelligent Performance Management
12. In some tenants, IPM sits inside the Planning cluster. If so, expand Planning (or your app name) and look for IPM / IPM Insights there.
13. Alternate route via Home cards
14. If you prefer, from Home:
15. Click a card labeled IPM, IPM Insights, or Intelligent Insights, or
16. Click your app’s cluster card (e.g., Planning) and then choose the IPM tile inside it.
17. Search if you don’t see it
18. Use the top bar Search (🔍). Type IPM, IPM Insights, or Intelligent, and select the matching entry.
19. Open the workspace
20. Click the IPM entry to launch the workspace. You should see the IPM landing screen (panels for Forecasts/Anomalies/Drivers may appear depending on configuration).
21. In the Navigator list or from the open workspace, mark it as a Favorite (★) if available, or Edit Home to pin an IPM card for one-click access.

Choose the required domain (Planning, Narrative Reporting, Consolidation, etc.)

From the Oracle EPM Home page, pick the application/domain where you want to run IPM-powered analysis. Common domains include Planning, Narrative Reporting, and Consolidation & Close (FCCS). The domain you choose determines which data, forms, models, and permissions IPM can use.

Why this matters

IPM analyzes the data and artifacts inside the domain you’re in. Choosing the wrong one leads to empty charts, missing models, or access errors.

Prerequisites

1. You can access the EPM Home page (Redwood UX).
2. Your user/role has visibility to the target domain (e.g., Planner/Power User for Planning, appropriate roles for Narrative Reporting or FCCS).
3. The domain has relevant data (time periods loaded, entities/scenarios, forms/reports/models as applicable).
4. Planning (EPBCS/Planning) → You need driver-based plans, forecasts, predictive models, variance/seasonality checks, scenario analysis, or to embed insights back into forms/dashboards.
5. Narrative Reporting (NR) → You’re producing management or board packs, need commentary + charts, and want IPM insights to explain trends in your narrative reports.
6. Consolidation & Close (FCCS) → You’re validating actuals, consolidation adjustments, intercompany, and you want anomaly checks or driver insights across consolidated data.

Detailed steps

1. Open the Application Cluster
2. On Home, locate the application card/cluster you need:
3. Planning (sometimes branded with your app name)
4. Narrative Reporting
5. Consolidation and Close (FCCS)
6. Click the card to enter the domain’s workspace.
7. Confirm You’re in the Right Domain
8. Check the header/app name and left Navigator entries.
9. For Planning you’ll see Forms/Data, Dashboards, Rules, etc.
10. For Narrative Reporting you’ll see Reports, Narratives, Packages.
11. For FCCS you’ll see Consolidation, Journals, Intercompany, Close Manager (labels may vary slightly).
12. Locate IPM/Intelligent Insights in This Domain
13. In many tenants, IPM appears as IPM / IPM Insights / Intelligent Insights within the chosen domain.
14. If you don’t see it:
15. Use the Search bar (type “IPM” or “Intelligent”).
16. Check the Navigator (≡).
17. Ask your admin if IPM is hidden in the navigation flow for that domain.
18. Check Data Context Before Running IPM
19. Verify Scenario/Version (e.g., Forecast FY2026, Plan Final).
20. Confirm Time Range (e.g., Jan–Dec 2025 loaded).
21. Ensure Entities and Drivers you care about are available.
22. In Planning, open a relevant form/dashboard first so IPM reads the right POV (point of view).
23. (Optional) Create/Select a Workspace or Report Pack
24. Planning: Open the planning dashboard or form that holds your measures (Revenue, Opex, etc.).
25. Narrative Reporting: Open the report package or report where you’ll embed insights.
26. FCCS: Open a consolidation view or report aligned to your close period.
27. Proceed to IPM Actions
28. Once you’re inside the right domain and context, launch IPM features (forecasting, anomalies, driver insights) from that domain’s IPM tile/panel.

Locate the Intelligent Insights / IPM tile and launch the feature.

From the Oracle EPM Home (Redwood UX), find the Intelligent Insights / IPM / IPM Insights tile and open it. Depending on your tenant’s navigation flow, the tile may sit on Home, inside a domain cluster (e.g., Planning), or under the Navigator (≡). Launching it takes you to the IPM landing screen where you can run forecasts, anomaly detection, and driver insights.

Prerequisites

1. You can sign in and reach the Home page.
2. Your role/group includes access to the target application and the IPM feature.
3. The IPM tile is visible in your navigation flow (your admin controls this).

Where the tile may appear (common placements)

1. Home → Card grid: A card labeled IPM, IPM Insights, or Intelligent Insights.
2. Home → Domain cluster (e.g., Planning, Consolidation & Close, Narrative Reporting): Open the cluster, then find the IPM tile inside.
3. Navigator (≡): Expand the left panel; look for IPM / IPM Insights / Intelligent Insights. Some tenants place it under Planning or a custom app name.
4. Search bar (🔍): Type IPM, IPM Insights, or Intelligent and pick the matching entry.

Detailed steps

1. Go to Home
2. After login, confirm you see the standard card layout (Redwood) with clusters like Planning, Dashboards, Data Exchange, Reports.
3. Scan for the tile on Home
4. If you see IPM / IPM Insights / Intelligent Insights, click it.
5. If not visible, open the Navigator (≡)
6. In the top-left, click ≡. Scroll the menu. Check under your domain (e.g., Planning) if needed.
7. Use global Search if needed
8. In the header search, enter IPM. Select the result that matches your target application.
9. Launch IPM
10. Click the tile/menu item. You’ll land on the IPM landing page. Typical sections:
11. Forecasts / Predict (time-series projections)
12. Anomaly Detection (outliers vs expected behavior)
13. Driver/Explainability (which factors drive results)
14. (Optional) Pin/Favorite for quick access
15. On Home: Edit (pencil) → pin the IPM card.
16. In Navigator: mark ★ Favorite (if enabled) to expose it on Home or Favorites.
17. Verify access
18. If you see “no access” or an empty page, you likely need the IPM duty role or navigation flow update. Share a screenshot with your admin for quick provisioning.

What to do immediately after launch (quick setup)

1. Set POV: Scenario/Version (Plan/Forecast/Actual), Entity/Region, Period range.
2. Pick a Target Metric: e.g., Total Revenue, Opex, GM%, Headcount.
3. Choose Analysis: Forecast / Anomalies / Driver Insights, then proceed to the next step in your guide.

Troubleshooting (fast)

1. Tile missing → Use Search; check Navigator; ask admin to unhide in Navigation Flow.
2. Access error → You have the tile but not the permissions; request IPM duty roles/data grants.
3. Blank panels → Wrong POV or no data; open the correct Planning form/dashboard first, or broaden the period/entity filters.

Upload or select the planning model or dataset for AI analysis.

Point IPM to the data it will analyze. You can either use an existing Planning model (cube/plan type + POV) or upload a prepared dataset (e.g., CSV/XLSX). Then you’ll map fields, validate rows, and save a reusable dataset for forecasting, anomaly detection, and driver insights.

Prerequisites (data readiness)

1. History depth: ≥ 18–24 months for time-series features (seasonality/trend).
2. Granularity: Monthly or weekly periods consistently populated.
3. Dimensions aligned: Account/Entity/Product/Scenario/Version/Period names match EPM metadata.
4. Clean signals: No duplicate keys (Entity+Account+Period), missing periods flagged/filled, outliers reviewed.
5. Drivers available (optional): Price, Units, Discount, FX, Campaigns, etc., in tidy columns.

Two ways to provide data

A) Select an existing Planning model

1. Choose Application & Plan Type
2. Open the Planning app (e.g., EPBCS) and pick the plan type/cube (e.g., Plan1, FS, Workforce).
3. Set POV (very important)
4. Scenario (Plan/Forecast/Actual), Version (Working/Final), Entity/Region, Period range (e.g., Jan-2023…Dec-2025).
5. Open a form or dashboard that reflects this POV so IPM reads the right slice.
6. Pick Target Metrics & Drivers
7. Target: e.g., Total Revenue, Opex, Headcount.
8. Optional drivers: Units, Price, Discount %, FX, Mktg Spend, etc.
9. Validate Preview
10. Confirm row counts, date coverage, and that numbers match the form. Fix security or filters if data looks sparse.
11. Save as “IPM Dataset”
12. Name clearly (e.g., IPM_Rev_US_East_Plan_Working_FY23-25) for reuse in later steps.

B) Upload a flat file (CSV/XLSX)

1. Prepare file
2. One row per period/entity/account; columns for Period, Account/Measure, Entity/Product, Value, plus Drivers.
3. Use ISO or EPM calendar labels (2024-01 or Jan-24) consistently.
4. Upload
5. Go to Data Integration / Data Exchange → Import Data (label may vary) and upload the file.
6. Map fields to dimensions
7. Map columns to Account/Entity/Product/Scenario/Version/Period/Measure.
8. Define time format and currency if required.
9. Run import & validate
10. Check load log, row counts, duplicates, missing members, or unmapped values.
11. Create IPM dataset
12. In IPM, select the imported source, choose Target metric and Drivers, set the time window, and save the dataset.

Quality checks before analysis

1. Coverage: No gaps in critical months; handle missing values (impute or exclude).
2. Outliers: Tag/winsorize known one-offs so forecasts don’t overreact.
3. Consistency: Same currency, same calendar; confirm FX handling if multi-currency.
4. Security: If preview is empty, you may lack data grants for the POV.
5. Name & version: Include app/domain, POV, and date window in the dataset name.
6. One metric per dataset (when learning): Keeps results interpretable.
7. Keep a “gold” dataset: A curated, validated slice you reuse across IPM steps.
8. Re-refresh after loads: Rebuild the dataset once new actuals are loaded.

Troubleshooting

1. “No rows / very few rows”: Wrong POV, missing role, or filters; widen Period/Entity; check security.
2. “Unmapped members”: Fix metadata or mapping rules; re-import.
3. Weird seasonality: Calendar mis-match or mixed entities; standardize period labels, split datasets.
4. Duplicate keys: De-duplicate by (Entity, Account, Period); aggregate if needed.

Execute automated forecasts based on historical data patterns

In this step, you initiate the automated forecasting engine within Oracle EPM to generate future performance projections based on past trends and business behavior. The system leverages historical datasets, seasonality, cyclical variations, and business driver patterns to produce predictive values for upcoming periods. This allows you to move beyond manual assumptions and rely on data-driven insights for planning and decision-making.

To begin, select the planning scenario or dataset you want to forecast, such as revenue, expenses, headcount, bookings, or product-level metrics. Choose the appropriate time range which includes sufficient history — ideally at least 18–24 months of clean, consistent data to help the model understand seasonal trends and structural patterns. Once the historical dataset is selected, launch the Predict / Forecast / Predictive Planning function in IPM (label may vary by environment).

The system automatically analyzes your historical series and applies statistical and machine-learning techniques to extrapolate future values. These typically include trend components, seasonal decomposition, smoothing techniques, and anomaly filtering. IPM may generate the primary forecast along with alternative model outputs, providing flexibility to review and compare.

After processing, the forecast results will appear in chart or table format. Pay attention to the predicted trend line, variance versus actuals, and confidence intervals that indicate uncertainty. Validate whether the projections align with business expectations, known market conditions, and upcoming business events such as product launches or campaigns.

If required, refine and re-run the forecast by adjusting input windows, excluding anomalies, or changing the model preference. Once satisfied, you can publish the forecast back into planning forms, use the result for scenario modeling, or export it to dashboards and narrative reports. This step ensures your planning process is both future-focused and data-driven, improving accuracy and strategic clarity.

Interpret the model-generated visuals to understand where performance is heading, how certain the projection is, and which patterns are driving it. You’ll use these charts to validate model quality, spot risk, and decide what to adjust in your plan.

What you’ll see & how to read it

1. Forecast Curve (the solid projection line)
2. Shows the model’s expected values for each future period.
3. Check shape (uptrend, flat, cyclical), turning points (inflections), and level shifts after structural changes (price lists, policy changes, new products).
4. Compare forecast vs Actuals/Plan/PY to judge reasonableness. If the curve overreacts to one-off spikes, consider excluding outliers or shortening the training window.
5. Confidence Interval Bands (shaded cone)
6. The band around the forecast (often 80%/95%) quantifies uncertainty.
7. Narrow band = stable, well-explained history; wide band = volatile or limited data.
8. Use the upper/lower bounds for risk-aware scenarios (e.g., staffing, inventory, cash). If bands explode after a certain horizon, cap your operational commitments there.
9. Trend Insights (drivers, seasonality, decomposition)
10. Breaks the series into trend (long-run direction), seasonality (repeatable patterns like Q4 spikes), and residuals (noise/anomalies).
11. Validate that seasonal peaks align with business reality (festivals, quarter-ends, launch cycles). Misaligned seasonality suggests wrong calendar, missing data, or mixed entities.

What to do with the insights

1. Sanity-check: If forecast diverges from known initiatives or constraints, adjust assumptions (price, volume, mix, ramp).
2. Scenario bounds: Use confidence limits to size best/worst-case capacity, budget buffers, and reorder points.
3. Refine the model:
4. Add/remove drivers (discounts, campaigns, FX) if residuals are high.
5. Backtest with a holdout period; review MAPE/MAE to confirm reliability.
6. Re-run after fixing outliers or extending history to capture seasonality (≥ 24 months ideal).

Troubleshooting quick guide

1. Flat forecast: Not enough variation or heavy smoothing → widen window, add drivers, reduce smoothing.
2. Wild swings: Outliers or structural breaks → trim anomalies, split pre/post event windows.
3. Bands too wide: Short history or noisy data → aggregate to quarters, add explanatory drivers, ensure clean loads.
4. Seasonality looks wrong: Check time granularity, calendar mapping, entity mix, and missing periods.

See outliers or unusual activity flagged by the system

In this step, you use IPM’s Anomaly Detection feature to identify unexpected patterns or deviations in your data that may require attention. Anomalies, also called outliers, represent values or trends that significantly differ from the expected range based on historical behavior, seasonality, and business drivers. These variations may signal risks, data issues, or emerging opportunities.

When you open the anomaly insights panel, IPM automatically scans your selected dataset and highlights periods or data points that behave outside the predicted trend. Each anomaly is visually marked and supported with a confidence score, making it easier to assess severity. You can drill into any flagged point to view its context—such as comparison with prior periods, target benchmarks, or forecast ranges.

Common examples include sudden revenue drops, cost spikes, abnormal headcount changes, unexpected expense variations, or major shifts in customer demand. After reviewing anomalies, validate whether they reflect real business events (like promotions or market changes) or potential data entry issues. This process ensures accuracy in planning and improves decision-making by drawing your attention to areas that may need adjustment or deeper investigation.

IPM identifies business drivers affecting performance

Use IPM’s Driver/Contribution/Explainability view to discover which variables (price, volume, discount, seasonality, campaigns, regions, etc.) most influence a target metric (Revenue, Margin, Cost, Bookings). You’ll set the context (POV), run the analysis, read the ranked drivers, and turn findings into actions. Driver insight explains the why behind trends and forecasts. It shows which levers move your results, by how much, and in which direction—so you can focus effort, refine assumptions, and build better scenarios.

Prerequisites

1. Data history: At least 12–24 time periods of clean data (more is better).
2. POV access: Proper role/data grants for the Entities, Scenarios, Versions, and Periods you’ll analyze.
3. Mapped drivers: Source drivers (price, units, channels, FX, headcount, campaigns, etc.) are loaded or linked to the target metric.
4. Model ready: If your tenant uses IPM’s explainable models (e.g., feature importance/SHAP), ensure the model has been trained/refreshed post-latest load.
5. Driver / Feature: A variable that may influence the target (e.g., Units Sold).
6. Contribution / Importance (%): How much that driver explains movements in the target over the window.
7. Direction: Positive (↑ target when driver ↑) or negative (↓ target when driver ↑).
8. Elasticity / Sensitivity: Expected change in target for a 1-unit or 1% move in the driver.

Detailed steps

1. Set the POV (context)
2. Choose Scenario/Version (e.g., Actuals FY2024, Forecast FY2025 R2), Time range (e.g., Jan-Sep 2025), and Entity/Brand/Region as needed.
3. Open Driver Insights
4. In IPM, select Driver Insights / Contributions / Explainability (label varies by tenant).
5. Pick the target metric
6. Examples: Total Revenue, Gross Margin %, Opex, Units.
7. If available, select the granularity (monthly/quarterly) and comparison (vs prior period, vs plan, vs prior year).
8. Choose analysis window
9. Typical: Last 12–18 months for stable insights.
10. Shorter windows emphasize recent shifts (campaigns, shocks); longer windows surface structural drivers (seasonality, macro).
11. Run / refresh analysis
12. Click Run, Refresh, or Explain. IPM trains or retrieves explainable metrics and shows a ranked driver list with contributions.
13. Read the ranked drivers
14. For each driver, review:
15. Contribution % (share of explained variance)
16. Direction (positive/negative)
17. Confidence / stability (if shown)
18. Example periods where the driver was most impactful (sparkline/drill)
19. Drill into a driver
20. Open the driver detail card to see:
21. Correlation over time, partial dependence (how target moves as driver changes), and segments (by product/region).
22. Segment to validate
23. Filter by Region, Product, Channel, Customer Tier to check if the driver is global or local.
24. Compare top vs bottom performers to see divergent effects.
25. What-if & assumptions
26. Use Adjust Driver (or pivot to a Planning form) to nudge a driver (e.g., +2% price, +10% units) and observe predicted target impact.
27. Capture the lift/drag to feed your Forecast or Scenario versions.
28. Capture the story
29. Click Export (image/CSV) or Add to Dashboard/Report.
30. Use Narrative to auto-generate a summary (if enabled), then refine with business context.
31. Close the loop
32. Update Planning assumptions (driver rates, seasonality factors, elasticity) and re-run IPM to confirm improvements.

How to read the results

1. High positive contribution (e.g., Units +42%): “Unit growth explains most of the revenue increase; each +1% units adds ~₹X Cr.”
2. High negative contribution (e.g., Discount Rate −18%): “Higher discounts are eroding revenue/margin; lowering discount by 0.5 pp could recover ~₹Y Cr.”
3. Seasonality (e.g., Q4 Seasonality +25%): “Year-end demand is a structural driver; pull campaigns into Q4 to amplify.”
4. FX or Commodity driver: “INR depreciation added +₹Z Cr to revenue; hedge exposure or adjust guidance.”

Best practices

1. Enough history: Target ≥ 24 months for stable seasonality/driver weights.
2. One change at a time: When testing what-ifs, adjust a single driver to isolate impact.
3. Watch multicollinearity: Highly correlated drivers (Price & Discount) can split or obscure contributions; simplify the driver set if needed.
4. Handle outliers: Large one-offs (extraordinary items) can distort importance—tag or exclude them in the analysis window.
5. Document assumptions: Record the elasticity you use in Planning so others can reproduce the scenario.

Troubleshooting

1. “No significant drivers found” → Window too short, too few history points, or drivers not mapped; expand time range and verify ETL/mapping.
2. All contributions look flat/near zero → Target is noisy or heavily constrained; smooth with period aggregation (Q instead of M) or add missing drivers.
3. Contradictory directions → Correlated drivers or regime shift; split the window (pre/post event) and re-run.
4. Stale model → Re-train/refresh after new data loads; check last run timestamp.
5. Permission errors / empty cards → Confirm role, data grants, and that the driver artifacts are visible in your POV.

Validate insights and adjust assumptions if needed

Review the AI-generated insights and predictions, confirm they make sense for your business scenario, and update assumptions (like growth rates, driver values, or timelines) if needed to ensure accurate and realistic results.

Save charts, export results, and share automated performance narratives

After reviewing insights in IPM, you can save charts, export the results, and share automated performance narratives with your team. This lets you quickly communicate key trends, forecasts, and anomalies using visual reports or generated narrative summaries, helping stakeholders understand the data and make decisions faster.

IPM automates performance analysis, making your planning faster and more accurate by using AI-powered forecasting, anomaly detection, and insight generation.