Consumption Forecasting Best Practices

Review key guidelines to get the most out of your consumption-based forecasts.

Available in: Lightning Experience

Available in: Enterprise, Unlimited with Sales, and in Agentforce 1 Sales Edition

Performance Considerations

If you have more than 14 million records for standard and custom objects, you incur higher data credit costs and pages load more slowly.
More frequent parallel page visits can lead to higher costs and latency.
In deep user hierarchies, higher-level users access a lot more data, which can result in varied runtime and performance.
Map your sales stages to forecast categories. To increase your forecast accuracy, adjust close probability based on historical data.
Cold cache can also affect performance.
To minimize the amount of data that’s accessed, select minimal forecasting periods, such as 3 or 6 months, rather than longer durations like 12, 18, or 24 months.

To improve forecasting performance, partition measure objects by engagement category and include an event date field.
To avoid unnecessary data loading, create multiple Data Lake Objects (DLOs) or Data Model Objects (DMOs) for large forecasts. We recommend that you base DLOs on specific products and use one DMO per forecast type.
Source each DMO from either a CRM data stream or a non-CRM data stream.
To prevent data errors and ambiguity, avoid combining CRM and non-CRM data streams within the same DMO.
To consolidate daily or weekly data into one entry per month, group raw data with Data 360 transforms, which reduces data volumes.
To further optimize data volume, combine multiple measures into one entry where possible.
To maintain a lean data footprint, keep only the data you need on main DMOs and regularly store older data in different DMOs.
To optimize data handling in Data Cloud One setup, use separate data spaces for different functions.

Did this article solve your issue?

Let us know so we can improve!