FAQs
  • 3 Minutes to read
  • Dark
    Light
  • PDF

FAQs

  • Dark
    Light
  • PDF

Article summary

What is Predict: Signals?

Predict: Signals is an AI engine that flags potential anomalies in your financial data. It is an assistant that gut-checks your data to find anomalies resulting from typos, formula-based errors, and misguided assumptions.


How does Predict: Signals work?

Predict: Signals trains on your historical actuals data and learns about your business trends, data seasonalities, and other statistical properties. It then uses this information to formulate a dynamic normal range. Data that falls outside of that range are then flagged as anomalies.


How is Predict helpful?

Predict works as an intelligent assistant for all your planning and analysis requirements by helping you save time and making you more efficient and accurate using the power of AI and your actual data.


How much data is required to use Predict: Signals?

Predict: Signals requires at least 3 years of historical actuals as a minimum for the AI engine to learn from your data.


Is our data safe and secure?

Yes, your data is absolutely safe. The Predict suite uses on-tenant training to evaluate your data. It does not use third-party services for training and your data does not leave your tenant.


What are the algorithms used?

Predict engine is built as a black-box AI engine for enhanced accuracy. We do not use a single algorithm but multiple algorithms, simulations, and mathematical models, and the outputs of all those are then stacked to generate the final results.


Can I use Predict: Signals in reports and templates?

Yes. Predict: Signals is available in both reports (where you can detect potential anomalies) and in templates (where you can potentially prevent errors from happening). The Predict engine looks at past months (closed periods) to generate signals based on actuals scenarios (e.g., consolidation reports) and it looks at future months (open periods) to generate signals on budget, forecast, and planned scenarios.


What are the risk categories and how are they generated?

Data points flagged by Predict: Signals are categorized as low, medium, or high risk depending on how far off they are from the normal range. Risk levels are generated by the AI engine and are extremely subjective to your data.


Can I resolve signals?

Yes. You can right-click on any signaled cell or set of cells and resolve them all. Once resolved, the anomaly flags for those GL values are then resolved across Planful.


What is Predict: Projections?

Predict: Projections helps create data-driven planning baselines using the power of AI/ML. It processes your historical data from actual scenarios within Planful to understand the trends, seasonalities, and statistical properties in the data. It then generates future projections using Planful’s proprietary AI/ML algorithms.


How does Planful: Projections help in financial planning?

Predict: Projections enables business users to quickly build realistic forecasts and budget scenarios using AI-driven financial forecasting informed by your historical financial data.


How do I generate AI-driven financial projections?

Simply right-click the line(s) you would like to forecast. You can create an entire AI scenario using Predict: Projections or use it in templates.


How accurate are the forecasts?

Accuracy is subjective to individual GLs or a group of GLs bundled together under a roll-up account for which you are searching for anomalies or generating projections. Planful has built our domain expertise into the Predict to help generate more reliable and accurate insights and forecasts.


Can I control the algorithms?

Not currently. But, very soon, you will have the power to choose from a supported list of algorithms for templates using Predict: Projections.


How long does it take to train the Predict engine?

It usually takes about 8-10 minutes for model training. It might take more or less time depending on your data volume. When Predict is enabled for the first time, it is recommended to perform the model training, which then automatically runs quarterly.



Was this article helpful?

What's Next