All Collections
Forecasts
Forecasting Guidelines & Best Practices
Forecasting Guidelines & Best Practices

Check out all of the tips and best practices behind forecasting with Trovata

David Taylor avatar
Written by David Taylor
Updated over a week ago

Forecasting Overview

Cash forecasts help predict your company's future cash flow and prepare for any cash shortages. Within Trovata there are two main building blocks, called Streams, that you can use to construct your Forecast.

The two Stream types are as follows:

  1. Data Stream (Consists of Machine Learning, Repeat History, and Manual Streams)

  2. Invoice Stream

Data Streams utilize historical transaction data to predict future cash flows in the Forecast model. These streams can be driven by machine learning, repeating historical data, or manual input.

Invoice Streams leverage open receivable and open payable invoices from your ERP system to predict future cash flow in the Forecast model.

General Forecasting Tips

Creating Forecasts

  • When creating a Forecast, use "All Accounts" when creating a company-wide Forecast model and use "Select Accounts" when creating a more specific, entity based Forecast model

  • Your forecast view will be impacted by your currency selection, which can be edited later at any time.

Data Streams

  • Selecting a Model Date Range determines which period of time, and in turn which transactions, are used by the algorithm to predict future cash flow

  • If the Model Date Range isn't selected, it will default to 3 months

  • There technically is no minimum or maximum when using the model date range selector, but the generally, the more data the better.

  • If there is no data for a certain period of time, the system will be pulling in 0's

  • Data Streams can be updated by clicking into the Stream, opening Settings, and updating the data source.

  • At least 100 transactions are needed in any Tag-Based Data Stream to resemble some form of accuracy

Forecast Manipulation

  • Use the Min and Max Value indicators to prevent the 4 models from going below or above a certain value

  • Data can be overwritten in any Stream

  • Any outflows will need to be entered in as negative values

  • Use the Editor to type in data directly, or use the Upload function to import data from an Excel document

  • Use the Weekend Manager to remove weekend data, or move it to a certain day/week

  • Use Factors to influence your Streams on a consistent basis into the future

  • The Growth +/- Factor is used for any % increase/decrease in cash values

  • The Shift Factor is used to push data backwards or forwards on the timeline

General Forecasting Best Practices/Troubleshooting

Machine Learning (ML) Data Stream Accuracy

  • Try setting the start of the Model Date Range on the day a transaction occurred and end on the day a transaction occurred

  • If your business experienced turbulence in a certain month that isn't representative of your cash flow, it may be helpful to not include that month in the Model Date Range

  • Generally, it's helpful to have a healthy range of data to capture in your Model Date Range (6 months or more)

  • However, it's also worth noting that if the Model Date Range is too broad, it may not pick up on certain trends

Did this answer your question?